Embedded Files
Amplifier Description:
The Pass Labs a40 is a high current, pure Class A (simple design, low distortion, lots of heat) stereo amplifier that uses bipolar junction transistors (BJTs) in a push-pull configuration for the output stage. My completed amp is biased just a little higher than spec and delivers approximately 50wpc into either an 8ohm or 4ohm load (measured with a scope). Each channel uses two pairs (two push, two pull) of darlington (essentially two BJT transistors in a single package) output transistors and dissipates approximately 100 watts all of the time. Total draw from the wall is approximately 190 watts (1.57A) for the completed stereo amp (bias adjusted for transistor substitution). Each channel has 112,000uF of capacitance, runs on +/-32v power supply rails, and is biased at about 3.15A, or 0.78A per output transistor. (The original design specifies +/-30v rails, and 0.8A bias per transistor, or 3.2A per channel.) At 0.78A bias, each of the transistors in the output stage will dissipate approximately 25 watts, thus heatsinking needs to be a minimum of about 1.0c/w for each transistor to limit thermal rise to 25c above ambient. In an interesting contrast between theory and reality, using larger than specified heatsinks ( rated at 0.67c/w each) and a bias level close to the target (0.78A per transistor) the thermal rise is relatively high at approximately 24c - looks like you can never have too much heatsinking... The total heat sinking for this amp is about 0.08375 c/w over a total radiating area of about 1728 square inches spread across 8 identical heat sinks. After running for approximately 1 hour, the heat sinks on each side of the amp run at a pretty steady 115F or so (depending upon room temperature) and the two channels have a DC offset of 2mV and 21mV. The completed dimensions of the amp are 18"w, 7.5"h, and 21"d.
What does it Sound Like?
I compared my newly constructed a40 to my Marantz MA500 monoblock amps (Class B, 125wpc) that I've been using since 1997. The Marantz amps are so much better than the inexpensive integrated amp that they replaced I was wondering how much of a difference I would notice this time. Both amps were compared using my Adcom GCD700 CD player and Atlantic Technology System 350 speakers. A pair of MA500 monoblocks cost the same amount of money as did my a40: approximately $600 for two channels (although it is certainly possible to build a stereo version of the a40 for much less - I exclusively used premium parts in my construction).
I spent a few hours listening to Dire Straits "Brothers in Arms" disk, switching wires back and forth between the amps. This is one of my favorite disks and happens to be well recorded to boot. The difference between the two amps was pretty clear.
Compared the to Marantz amps, the collection of instruments sounded much more "diffuse" and spread out with the a40. Mark Knopfler's voice also sounded much more clear, much more focused in the center between the speakers, and much louder (despite matching the levels of the amps with my SPL meter and a test CD beforehand) - interestingly, the musical instruments themselves did not sound louder. There seemed to be a greater depth separation between the vocals and the music. Switching back to the Marantz amps, it seemed as if the sound stage was compressed both front to back and side to side. Mark's voice sounded "flat" and less distinguished in point of origin from the music. Moving back to the a40, it was as if the music had just popped back into 3-D: Mark's voice was much more prominent, much more clear, and was projected at a location directly between and several feet in front of the speakers. The music was more perceptibly originating from behind the speakers and sounded more spread out to the left and right than the physical spacing of the speakers. Overall, it sounded as if the a40 amp was also "cleaner" and more realistic sounding than the MA500s. Listening to more of the Dire Straits CD, it sounded as if the music through the a40 had more midrange and more "warmth" than the same music through the MA500s. This may have been what contributed to Mark's voice sounding louder through the a40, while the music appeared to be at the same level.
I was really surprised at the perceived warmth of the a40 amp when it was compared to the MA500s. On a web forum, Nelson recently described the sound of his amps: ".. if you listen to the amps representing my progression as a designer, you will find a trend from accurate and analytical to warm and romantic." Since the design of the a40 is from 1978, I had expected it to lean toward the "accurate and analytical" end of the spectrum. Additionally, since the MA500s have been very highly acclaimed by the audio industry and are described by many as being warm and smooth sounding amps, I was expecting the a40 to sound "thin" in comparison. In reality, the a40 made the MA500s sound thin and flat. Overall, I am very pleased by the sound of my new amp, and very happy to get rid of the hum that crept in while installing the parts into the chassis.
I used Mark Knopfler's "Sailing to Philadelphia" CD to compare the Marantz & AT system to the HATT & a40 system. Cueing up the disk with the Marantz & AT, the sound was very familiar. Marks voice sounded very nice, just as I had become accustomed to. The speakers flank my 65" RPTV with about 18 of breathing room on each side of the TV, stand out from the screen of the TV by about a foot, and are toed in so that they converge just behind the listening position on the couch. (I know, its not ideal to have a big television right between my speakers, but its the best physical placement for now...) With this setup, there is no change in either the sound stage or imaging between standing or sitting positions, but both quite clearly collapse by moving as little as one foot to the left or the right while standing.
Switching to the a40 & HATT combination, there was a tremendous difference! The sound stage extended far to the left and right of the speakers. Sounds did not seem to emanate directly from the speakers as with the previous arrangement. Rather, the entire sound field was much more diffuse and spread out. Mark's voice was very noticeably "cleaner" and "clearer" - it sounded much more like he was standing where my TV was and actually singing in person! Standing up or sitting down made no noticeable difference in the sound stage or imaging, just as before. However, the sound field and imaging remained intact when I moved as much as 3 or 4 feet in either direction! The instruments seemed to be located far behind the physical position of the speakers, while Mark sounded as if he were closer to me than the physical positioning of the speakers would indicate. The sound was so much improved that I was very surprised!
Previously, I had compared the HATT speakers with the Atlantic Technologies using the Marantz amp and preamp. The HATTs were noticeably better sounding than the Atlantic Technologies. Similarly, on a different occasion, I had compared the MA500 amps to the a40 amp using the Atlantic Technology speakers. The difference between the amps was also very evident - with the a40 handily outclassing the fine MA500s. Thus, on their own, the HATT speakers clearly sounded better than the Atlantic Technology speakers and the a40 amp clearly sounded better than the MA500s. But the combination of the a40 and the HATTs brought things to an entirely new level!
Clarity and musicality are greatly increased by the a40 and HATT combination - listening to music is much more fun and enjoyable. Switching back to the Marantz and AT speakers, they sound considerably more muffled and much less three dimensional!
Yes, I know, I can hear you now "but you only listened to one CD, and it was pop music!" Someone else is surely saying "your listening test wasn't blind!" I'm aware of the fact that I may be biased (sorry for the pun) toward my own work, I'm also aware of the criticisms of blind testing... Perhaps the most interesting observation is that these differences were also noticed by my wife who is not nearly as wrapped up in audio as I am. While she was not able to tell at any given time which amp we were listening to, she could clearly tell which one it was when I switched amps. The point is that on music that I am very familiar with the improvements were clear. The same was true for a host of other CDs - though not all due to differences in the original recording quality.
After having time to spin a few more CDs, I'm happy to say that my impressions of this amp continue to improve. It was not just Dire Straits that sounded so much better, but Peter Gabriel, Enya, Sade, Nancy Wilson, Mozart, Beethoven, etc. Each new disk seemed to have more "life" than was apparent through the a40's store-bought relative. I've really gotten to have a great deal of respect and admiration for those shiny little 5" disks! Its really a shame that there are so few people who actually get to experience the entirety of the musical experience that is hiding of each of the disks that they own. Its absolutely amazing what a good subwoofer, an honest set of speakers, and good amplification can sound like! Once exposed, you, too, will thirst for more! As a side note, MP3 files just don't have the ability to do ANY amp/speaker combination justice. MP3 files us a form of digital compression that literally THROWS AWAY data (called lossy compression). While the resulting MP3 file may sound similar to the original recording on inexpensive equipment, MP3 playback on a high quality audio system will quickly reveal the flaws in the compression process.
I am very curious to hear what some of the more recent designs such as the Aleph series (which are from the mid 1990's) and the new Aleph-X design sound like since they are supposed to fall more toward the "warm and romantic" end of the spectrum, sounding more like tube amps than traditional solid state amps. If you are interested in building an Aleph amp (Class A, singled ended, mosfet amp) check out Mark Finnis' project notes and artwork for the PCBs or visit the Pass Labs Forum at DIYAUDIO. Nelson himself has been very supportive and continues to make excellent contributions to the DIY community!
Some General Construction Tips:
Yes, many of these were learned to hard way, so the reason for this section is to try to prevent you from going through the same frustrations that I did! In no particular order, here are a few things I've learned by working on this project off and on for over 2 years.
1) Measure each and every part prior to including it in your circuit. Mistakes happen at many different levels. In one case, I was sent resistors that measured 220k ohms when I ordered 220 ohm resistors. Also, in many cases, it is useful to match parts across a stereo amp or match components in the output stage with one another. Another benefit of doing this is that it will decrease the likelihood of making a dumb mistake like soldering the wrong part value into your circuit.
2) Make changes to only one channel at a time when working on a multi-channel amplifier! Often times, a change that you intend for the better results in the introduction of a problem. This is especially true when moving wires inside your chassis in order to reduce or eliminate hum. Other changes, if improperly applied, have the potential to let the smoke escape from vital components of your amp - its far better to toast one channel rather all of them at once! As a parallel to this rule, power up only one channel at a time after making a change...
3) Be especially careful when soldering to avoid cold solder joints. A cold solder joint occurs when the materials that you are soldering together have not yet reached the melting point of the solder, thus you have a hot blob of solder on a cold piece of metal. The result is a solder joint that is neither electrically nor mechanically solid. A cold solder joint can thus cause a great deal of trouble because when visually inspected, it looks good. When the component itself is measured with a meter, it measures good. But, the connection is intermittent at best. I've been soldering for years and have never had a problem with this before, but it caused some frustration in fixing a problem with this amp. A cold solder joint on one of my resistors resulted in running the full rail voltage to the speaker outputs which would have immediately fried any speaker! Ironically, I suspect the cold solder joint resulted from the use of a soldering heatsink that I had clipped to the leg of the resistor to prevent the resistor itself from overheating while I was soldering it. This problem is easily solved by carefully and methodically re-melting each of your solder joints with the tip of your soldering iron and then letting them cool again.
4) Power supply transformers, power supply capacitors, and rectifiers should all be chosen to exceed your anticipated or design requirements! For Class A amplifiers, Nelson Pass recommends a minimum VA rating for the transformer of 7.5 times the output power of each channel. Thus, if you have a 100wpc Class A mono amplifier, you need a minimum of a 750VA transformer - and more is almost always better in this case. For a given application, a larger transformer will provide a more stable source of power than will a smaller transformer - the voltage will droop less under the load of the amp, it will generate less heat, and is less prone to mechanical resonance under load. Power supply capacitors should be able to comfortably handle the voltage of the circuit. If you have 30v rails, aim for 50v or higher caps. Choosing 35v caps may not provide enough wiggle room should you experience power surges or if they are simply off tolerance - many caps are off tolerance by as much as plus or minus 20%. Thus, pick your caps so that they are comfortably beyond the 20% mark. Also, if the design calls for 20,000uF, its hard to go wrong (in Class A solid state design) by choosing 40,000uF (or higher) caps. Similarly, choose your rectifier diodes such that they can safely handle many times the voltage and current requirements that your design requires - especially if you are planning to really stress them out at turn on by using super large capacitors in your power supply. Although this amp draws only 1.6A when it is running, the current draw surges near 7.5A when the amp is first powered up. This causes lights in the room to momentarily dim and a temporary hum can be heard as the transformer comes to life. A bridge rectifier that is rated at 600V and 35A is not at all out of line - power up is fairly stressful. The caps that I have for my Aleph-X amp are 220,000uf each which will certainly stress any transformer and rectifier upon power up. When entering into the land of big transformers and huge caps (220,000uF per rail qualifies as huge) its time to start looking into "Soft Start" circuit or at least consider placing a thermistor in series with the transformer main.
5) Heatsinking: It seems that due to a number of thermal transfer inefficiencies you need to incorporate a "fudge factor" to your heat sink calculations. Much of the math behind calculating junction to case, case to heatsink, and heatsink to air conductance rates is based upon ideal, theoretical circumstances - circumstances which are never achieved in practice. This conclusion is based upon two examples: my a40 amp described here, and the Pass Labs Aleph2 monoblock amp.
Example 1: For my a40 amp, the article calls for heatsinking rated at .25 c/w for each channel in order to limit the theoretical temperature rise of the heat sink to 25c while dissipating 100watts for that channel. Using the formula of 25c temp rise / (c/w rating of the heat sink) in this case yields 25/0.25 = 100w. Thus, it looks like 0.25c/w of heatsinking per channel hits our expected heatsinking demand exactly: we can dissipate 100w of power while limiting thermal rise to 25c. Multiplying this by 4 (because I used 4 heatsinks per channel) dictates that we need 4 heatsinks, each rated at 1.0c/w - one heatsink per output transistor. However, for my completed amp, I used heatsinks rated at 0.67 for each output transistor and still achieve a 23 to 25c temperature rise. Thus, theory calls for 1.0c/w per transistor, while practice shows us that 0.67 is really necessary. Thus, we need to derate the capability of the heatsink to approximately 70% of its claimed dissipation.
Example 2: The Pass Labs Aleph2 monoblock delivers 100wpc into an 8 ohm load and dissipates 300 watts of heat all of the time. Nelson Pass indicates that the Aleph2 requires heatsinking of 0.06c/w per monoblock. Using the theoretical calculation dissipating 300 watts of power should require 0.083 c/w worth of heatsinking to limit thermal rise of the heatsinks to 25c above ambient. Derating the theoretical result of 0.083c/w to 75% provides us with a more realistic figure of 0.0625c/w - much closer to the target value of 0.06c/w indicated by Nelson.
So, for the conclusion: Perform your calculation of the power that you need to dissipate, determine the appropriate sized heatsink, then derate its dissipation to 75% of the calculated dissipation rating. This is the real size of a heatsink that you need - unfortunately, you always need more heatsinking that you think!
Ongoing Construction Notes:
Disclaimer: I am not an expert and, I suspect, neither are you (or you wouldn't be reading this). This project involves electrical wiring that connects to the 120v house mains. These voltages are lethal and you should take all appropriate precautions. If you don't know what these precautions are, be sure to take the time to learn before going any further!
When your parts arrive, be sure to measure each and every resistor and capacitor that you have received! Several times in the past I have received resistors of different value than what I had ordered. For example, I had ordered 220 ohm resistors from one source and received 220k ohm resistors marked as 220 ohms instead! I have exclusively gathered "exotic" parts for the construction including Black Gate FK Caps, Mills non-inductive wire wound resistors for the output stage, and Shinkoh 1/2w Tantalum film resistors throughout. To date, parts total $561; however, I did not purchase all of the parts listed below (nylon standoffs, EIC power filter, Bergquist sil-pads - many of these items were obtained for free as samples, etc). Next up is prototyping the chassis layout on wood. I have not yet decided whether I will use an aluminum chassis from Par-Metal or build a chassis out of red oak to match existing wood work in my house. As I have many other projects to work on, assembly progress will likely be slow.
I finally began to populate my circuit boards that I ordered from AudioXpress. I don't know how much of a difference it actually makes, but before soldering components onto the board, I cleaned the leads of each component with 0000 steel wool. This made a very noticeable difference in the shine of the wire lead, so I assume it was a useful step in the process. In addition to simply verifying the value of each resistor, capacitor, etc, I matched each component across channels as closely as possible. Also, while stuffing my PCBs I discovered that I was unable to determine the pin out of the 2N5248 FET devices I had obtained from Ack using the link below. After further investigation, it seems that over the years there has been several different pin outs for the 2N5248 N Channel J FET in the plastic TO-92 casing. Consider the following examples:
1) One source lists pin 1=S, 2=G, 3=D
2) NTE device NTE312 (a 2N5248 equivalent) lists pin 1=G, 2=S, 3=D
3) The original a40 article (figure 10) lists pin 1=S, 2=D, 3=G
So now we have three different pin outs from three different sources, all for the "same" FET. There are three solutions out of this little snafu: 1) purchase parts from a vendor that clearly identifies the manufacturer of the part (such as Mouser) AND who also supplies a spec sheet for it, or 2) use and ohm meter to determine the pin out of the device, or 3) test the FET as indicated in Nelsons' article about DIY Op Amps.
To identify the pins with an ohm meter, follow the advice from Mike Rothacher (a DIYAudio member): "Generally, the Source to Drain resistance should indicate about 150-300 Ohms resistance in both the directions and the remaining terminal should show a diode's behavior with these two pins. That is, high impedance in one direction and low impedance in the opposite direction." Brian Segura was able to verify the accuracy of using this method by testing the NTE312 FET that he received from Mouser.com. See this image for his measuring results. The top set of measurements are from the NTE312, while the bottom set are the 2N5248 from Ack Electronics. I used my meter with the above procedure to determine the pin out of my FETs and had no problem with them at all!
In my haste to hook everything up, however, I mixed up my N-channel and P-channel output transistors! Duh, rookie mistake, I know... This little mistake immediately smoked R15 on the circuit board, blew R16 through R19, and cooked Q6. After replacing these, and putting the output transistors in their proper location, the amplifier produced very nice sound, but also included an annoying little hum. Good thing I only powered one channel at a time!
Also, Nelson recommends adding heatsinks to Q3, Q4, and Q5. If you are having trouble finding heatsinks for TO-92 devices (I wasn't able to find any..), you can make you own with strips of aluminum or copper. Cut the strips approximately 1/4" wide and 1" long and wrap the strip around the transistor leaving a long ear, it would look like a letter "b". I used needle nose pliers in order to properly bend the strips without disfiguring the transistors. Then use some thermally conductive adhesive like JB Weld (available at auto parts stores, Walmart, and online) to hold your new heatsink in place. After about an hour of operation, these heatsinks reach about 40c (just above body temperature), but no warmer. This picture is an ariel view of my DIY heatsink, the grey blob in the middle is the JB-Weld that holds the copper to the transistor.
Tracking down this hum problem has been pretty time consuming. Working on advice from Nelson Pass, I first tried to determine what type of problem it was. For this Nelson makes the following recommendations:
"When you run into this sort of stuff, you need to isolate the problem down to [one of] the following categories: 1) oscillation: high frequency buzzing, none of the other approaches helps: play with frequency compensation, look at output with a scope. 2) ground loop: If the problem doesn't go away when you disconnect the source and short the inputs, it's not a ground loop, probably. 3) transformer pickup: Move the transformer / bridge / cap assembly away physically, and the problem goes away. 4) poor ripple rejection: CLC or CRC filter in the supply fixes this."
Well, it was pretty easy to eliminate ground loop as one of the potential causes of the problem. Strangely, following Nelson's advice, shorting the input to ground increased the hum, rather than reducing it! Moving various wires around inside the chassis didn't seem to have much effect at all. So it seemed, I was not dealing with ground loop. Next, I removed the power supply from the chassis. Yep, I yanked out the transformer, bridges, and the filter caps. I attached them to a separate board and used wires with alligator clips to run the power back to the amp. I also added an RC snubber circuit to the bridge rectifiers. Each diode was bypassed with a 100ohm resistor and a 0.047uF capacitor to help reduce the RF emissions from the diodes as they turn on and off. This was an improvement, but I still had a quiet hum emanating from the woofer when the input was shorted. The tweeter had a very quiet hiss that was present, but only if you pressed your ear directly against the driver. The hum from the woofer was much better than when the power supply was inside the chassis, but still noticeably present. Next, I tried fiddling with the power supply itself. In its original configuration, the transformer was connected to the rectifiers, and then to the 24,000uF caps, one per rail. I changed the configuration to have a CRC filter: the bridge fed two 24,000uF caps (one for each rail for one channel) and then I added a 0.34 ohm resistor (I simply paralleled to of the 0.68 ohm wirewound resistors left over from the output stage) to one cap and ran it into a second 24,000uF cap and then used this to power the amp. So now I was using all 4 caps for a single channel: 2 on the positive rail and 2 on the negative rail. Upon power up and connection to my CD player, the amp was absolutely dead silent (until I pressed Play!). No hum, no buzz, not even a hiss from the tweeter with my ear pressed up against it. No turn-on or turn-off thumps from the speaker - nothing by silence!
So now that the problem has been identified (I think...), I have the fun task of re-installing the power supply back into the chassis while trying to keep it from producing a hum again! I have reconfigured the power supply to put the bridges closer to the transformer and keep them as well as the caps further away from the front end PCBs. Another trick I am trying is to put some shielding between the power supply and the rest of the circuit. I've got some copper flashing that I now need to cut down and install to cover my wiring. I'm also not sure if the amp really needs CRC filtering, or if my surplus acquired caps are just beyond their useful life. One thing is sure, though, I don't have room for eight 2" diameter 6" tall caps in my chassis. So it looks like I will end up with a CRC filter with 10,000 or 15,000 uF then the resistors, followed by 20,000 or 30,000 uF of new caps...
Oh, by the way, using the Motorola (now On-Semi) MJ11015 & MJ11016 transistors in place of the originally specified Lambdas results in a bias level that is just little bit low. To bump up the bias level (and the resulting heat generation...) I increased R11 from 4k75 to 5k6. To reach the target bias level (see discussion below about bias) you should be able to measure approximately 0.55v across any of the source resistors on the output transistors. After the amp has warmed up, I measure approximately 0.50v with the a 4k75 resistor for R11 - after changing this to 5k6 the dV across each of the source resistors averaged 0.60V. Temperature rise on my heatsinks (one transistor per 0.67c/w heatsink, using 0.20c/w K-10 sil-pad) is about 27c with the higher value resistor.
The transformer has now been returned to the chassis and there is no hum. I cut a band of copper flashing just a little longer than the circumference of the toriod and made it about 1.5 times taller than the transformer. I also left 3 or 4 tabs along the bottom of the flashing and drilled holes through them in order to attach the copper band to the chassis. The toriod sits in the middle of the copper ring and the bridge rectifiers sit on top of the toriod When I power up the amp, the transformer no longer produces a hum through the speakers - one problem solved (best part is the cost of this fix: less than $5). I installed an RC snubbing network across each of the diodes in my bridge rectifiers. There is a very detailed article by the name of Calculating Optimum Snubbers that indicates how to calculate the ideal values of R and C for each particular transformer and bridge combination. Performing calculations according to the article for my components resulted in target resistance of 100ohms and capacitance of about 0.047uF - pretty close to the generic recommendation of using a 100ohm resistor and a capacitor in the range of 0.33 to 0.47uF. Adding these snubbers to the bridges greatly reduced the high frequency buzz that was emanating from my tweeters. Now, there is just one problem left to solve: a buzz coming from the woofers.
Using the surplus caps that I obtained from MECI now seem problematic. If I use only a single cap on each power supply rail, I get a significant amount of buzz through the woofers - most likely a result of power supply ripple (I don't have a 'scope). If I place two caps in series with a 0.33ohm resistor between them (traditional CRC filter), the hum disappears. Also, if I simply parallel two caps without the resistor, the buzz disappears. Since the amp operates on 30v rails and draws only 1.8 amps per rail, 24,000uF caps should be plenty to remove power supply ripple. Its not too hard to conclude that the surplus (and heavily used) caps I purchased are dried out and need to be replaced. New ones are on their way, we'll see how they work. I ordered a set of four caps rated at 65v with 56,000uF of capacitance, one for each rail... Late Edit: When ordering your caps, especially if they are used, see if you can learn when they were manufactured before you purchase them! The date code is a series of 4 digits and will likely include a suffix of one or more letters. In the example below, the date code is the last line of printing on the cap and reads: "8008L" The first two digits indicate the year of manufacture (1980) and the last two digits indicate the week (8th week of 1980). The L indicates which plant or facility actually manufactured this particular cap. In general, I would be suspicious of any cap that is more than about 10 years. Many that are older are still fine, but the chances of getting an old and dried up capacitor (like I did) increase as it gets older. The original caps I got had a similar date code: from 1980.
I also discovered that my shielded input signal wire was the source of lots of buzz... As is turned out, the wire was mounted to the side walls of the amp and ran from the front to the back of the chassis where it picked up interference from the power transformer! Using well shielded coaxial wire did not prevent the interference problems that I thought it would! So, I replaced the coaxial wire with three strands of insulated 30ga silver wire that were braided together and made sure that they were located safely behind my diy transformer shield. And so another source of buzz has been eliminated!
My new caps arrived and they have indeed solved the ripple problem from my power supply - still don't have a 'scope, but the audible effects of my surplus caps are gone. The also have a manufacturing date code that indicates they were made in 1995 - much better! All components are now installed back into the chassis and there are no ill effects - the speakers are silent when no signal is present (although the amp looks like a rat's nest of wires). My next task is to shorten all of the wires and tidy up a bit and then add the front and top to my chassis. I haven't yet performed any measurements, but the new caps have increased the bass response and seem to have affected the bias level just slightly. I'll have to remeasure the temperature heatsink temperature and bias current but it seems that the amp is running just a little hotter to the touch with the new caps. The heatsinks are still cool enough to leave you hand on them for an indefinite period of time. Also, the bridge rectifiers are a good deal warmer than they were previously - the temperature of their DIY heatsink is very close to that of the output transistor heatsinks (near 40c). According to an IRF data sheet, an average 600v 35A bridge can handle temperatures up to about 130c at a load of about 5 amps. Running the bridge at 40c seems to be well within the Safe Operating Area of the rectifier...
On a side note, this amp is not happy using my DIY Gomer Speaker Cables. When both channels are hooked to the amp, the speakers emit a fairly loud hum, most likely a resonance developed as a result of the fairly high capacitance and inductance presented by these wires (approximately 5.3H and 4.6nF for 23 foot lengths). Sounds like the characteristic "motorboating" sound that is so often described with these types of cables (very similar in construction to Kimber Kables). Looks like I'll be using more "traditional" speakers wire with this amp...
One final tweak was added to the power supply. I separated the power supply going to the input stage (on the PCB) from the rest of the power supply (that feeds the high current output stage) by placing a 1N4002 (standard recovery, 100V, 1A - available at DigiKey) diode and a 2,200uF 50V cap between the main filter cap and the PCB. Also, each large filter cap is bypassed with an AXON 4.7uF film cap. There were two effects I noticed after doing this: First, there was a slight turn-on, turn-off thump that went to the woofer before the bypass cap was added. It is now gone. Also, the amount of hum coming from the speakers is reduced now that the input stage is better isolated from the power supply for the output stage. I have not yet given a critical listen to the amp, so I can't directly comment on any before/after sound quality differences resulting from these changes. The addition of the diode and the extra cap drops the power supply to the input stage by about 0.6V so now the output stage runs on about 32.7V while the input stage runs on about 32.1V. From what I can tell, this does not cause any problem at all for the amp.
The amp is nearly completed. All of the electrical work for the circuit is in place, all that remains is to add a front and top to the chassis and hook up two blue LEDs that will serve as power indicators- one for each channel. After months of construction, I can reliably pass on several pieces of advice: First, when hooking things up, testing the power supply, testing the amplifier circuit, determining routing for wires, etc, WORK ON ONLY ONE CHANNEL AT A TIME! If you manage to hook something up incorrectly (as I did) you only lose one channel - not both (pfew!). Second, its frustrating to hook things up with really long wires that loop around and look sloppy (but the amp works well) only later to shorten the wires to clean up the chassis and find that you induced hum into the circuit. I've spent countless hours moving wires to reduce hum. In my particular instance, I found it beneficial to 1) shield the transformer with a copper band, and 2) keep any power supply wires (AC to transformer, transformer to rectifiers, rectifiers to caps) as far away as possible from your signal wires. When the amp was almost done, I had somehow induced hum (again!). The fix was to disconnect the wire that runs from the rectifier to the caps, move it one inch farther away from the signal wires and reconnect it. Its surprising how much impact such a small item can make - its also very frustrating until you track it down! Finally, when soldering your components onto your PCB, it is handy to have a soldering heatsink clipped to the leg of the component you are working on. This will help prevent the heat from your hot iron from damaging the component if for some reason it takes a little while longer to get a good solder joint than you thought it would.
The circuit boards have been secured to the chassis, the front panel has been attached, and the blue LEDs have been installed. Even using a resistor to run the LEDs at approximately 1/4 of their full output level, they are very bright and certainly capable of "spot welding your retina from across the room" as someone recently put it! It looks like I'll have to install a bigger resistor to tame those tiny - yet excessively bright - spot lights! Lastly, once I install the top cover it will be done (pfew)! Talk about your slow progress - I started gathering the first parts for this amp in December of 2000! I'll post a few pictures by week's end.
I have finally taken and scanned a few pictures. The completed dimensions of the amp are 18"w, 7.5"h, and 21"d. I also ended up changing the resistors on those LEDs. They run on 4 vDC and have a max rating of 20mA so I used [(30v-4v)/0.02A = 1300 ohms] 2,800 ohm resistors - effectively running them at half of their maximum current. The result was blinding! I am now running them at approximately 0.6mA (44,000 ohms) and they are still pretty bright, but much better! I may still change this value to a 100k ohm resistor to bring the brightness level down again...
The signal input wire is solid core 30ga silver in a teflon insulator, and the low-current wiring from the PCB to the output transistors is composed of 8 strands of the same 30ga silver wire. I used Wonder solder which has a high silver content and is also eutectic (discrete melting point). The speaker level output is high purity 12ga stranded copper wire - I don't have any silver wire of sufficient gauge for this connection. Looking at the picture of the completed amp, you can see that there is some left-over room inside the chassis - it was designed to fit the heatsinks I was able to find. Building this amp again would likely result in a different configuration to match a different set of heatsinks.
Looking at the picture of the front of the amps, you can see the copper shield that surrounds the transformer as well as the bridge I used to connect the grounds from each channel to AC ground - I used Nelson's approach from the latest version of the Zen amp to help prevent ground loops. The photo of the rear of the amp is from a while ago before the amp was completed, thus the wires sticking out all over the place! I also have a little staining to do the clean up the overall appearance of the amp. The best part of the chassis was its cost: $1, a local furniture shop was having a sale on left over parts and I picked up a 12" wide bed rail for $1 and cut it down. Its a high-grade plywood with a very nice veneer on it that was already finished!
In order to help minimize the possibility of hooking things up incorrectly, I tried to use a standard wiring convention: red wires carry positive voltages, white or black wires carry negative voltages, and green wires all lead back to ground (the center taps from the transformer). Believe it or not, it helps prevent dumb mistakes when you have a mess of wires in your hand!
All in all, its most likely not the most tidy arrangement for an amplifier that you've ever seen, but it is quiet (no hum from the speakers) and it sounds very nice, certainly better than similarly priced commercial gear.
A few final tweaks to the amp include adding a top to the chassis and simultaneously allowing for better ventilation with the chassis top installed. First, the top of the chassis: This is simply another board that was attached to the rear panel of the amp with a set of hinges so that I could easily lift the lid to do any further work (a hood, if you will). Next, with the lid in place, ventilation through the chassis needed to be improved. I put a set of self-adhesive felt pads on the underside of the lid near the front panel. The felt pads together with the gap created by installing the hinges in the rear of the amp resulted in a 1/4 inch air gap all around the lid. Then I drilled a series of 1/2 inch holes in the bottom panel of the chassis to create some air flow. Since the transformer and bridge rectifiers generate most of the heat inside the chassis, I drilled a series of holes between the transformer and the copper RFI shield I installed. Be careful doing this after the transformer is installed, as nicking the transformer with the drill bit will likely create a short or other dangerous situation! Finally, since the bridge rectifiers ran a bit on the warm side (a little hotter than the heatsinks), I replaced each 3 inch aluminum "heatsink" strip with a 5 inch strip. The larger surface area together with increased airflow through the chassis keeps the bridges a little cooler now.
So that's the story of my a40. Excuse me while and listen to my latest creation…
One Final Update:
After 4 years of use, the EIC power entry unit from that is featured in the third picture from the left at the top of this page has started creating problems. It is an integrated power input, power filter, and switch. Its the switch part that has created the problem, it will no longer turn off (which, I guess, is better than never turning on). Something in the rocker mechanism has broken rendering the switch always on. On my next amps (the Aleph-X, see link above) I have already chosen a set of more sturdy steel toggle switches (rated for 10A at 240v). I will just use a "standard" EIC power entry without the switch and then put the switch right above it.
The Parts List:
After a lengthy search for vendors for appropriate parts, I assembled the list below and slowly began collecting parts in December of 2000. The resulting list below contains nearly everything (see exceptions) needed for building the Pass Labs a40 amplifier. Numbers on the left (in parentheses) represent the quantity of each item that is required for a complete a40 stereo amp and the corresponding part number in the amplifier or power supply schematics. Dollar figures (US$) at the end of each line represent the extended cost of each item (quantity * unit cost). For each part, I have indicated the specific part/stock number from the respective vendor. If the links fail to work, go to the vendor web site and search on the indicated part number. While there may be other vendors that carry these parts, these are the ones I have found that allowed me to reduced the total number of vendors and minimize the number of orders to pay shipping costs on. In some cases, certain vendors have a minimum order value. Where this occurs, I simply purchased a few additional parts that will be useful elsewhere. Parts are grouped by vendor and subtotals are estimated.
(1: T1) 600VA toriod with 4 secondaries, 24v, 6.25A each - Victoria Magnetics $95 [Update - Victoria Magnetics is no longer in buisness] Article specifies a dual 44v (22, 0, 22) center tapped E-I transformer, if you use toriodal transformers (smaller, lighter, less magnetic radiation than E-I type) you will need *2* stock toriod transformers, each with dual secondaries to create the center taps. John Snowden at Victoria Magnetics was kind enough to custom wind a single toriod with 4 secondiares for me. After rectification and at full load, this transformer provides between 32.3 and 32.7 volts DC (effective multiplier is thus 1.34, a little less than the theoretical 1.414 multiplier). After about 3 hours, the transformer gets warm, but is certainly cooler than the heat sinks (which exhibit a 23c temperature rise). John provides excellent customer support in addition to his fast service. Overall, I would say that 300VA per channel is an absolute minimum. Now that my amp is completed and I've been playing with different bias levels, I'm wishing that I had a 700VA or even 800VA transformer to allow some more freedom...
(1) Par-Metal custom chassis - 12 Series 12w 20d 6h ~$125 (dimensions accommodate heatsinks below). Rather than use a metal chassis, I constructed one out of wood. On a side note, I have increasingly become aware of a large number of people who advocate using wood for first-time chassis construction. After several months of working with my amp, I would agree!
(2) a40 printed circuit board - Audio Xpress $12 (this board contains a trace error, be sure to read the paperwork from Audio Xpress. Alternately, figure 10 in the a40 PDF file on the PassDIY web site shows the proper circuit board configuration. The area to pay attention to is the top left corner of the board as shown in figure 10.) Have a look at the correction document that came from Old Colony when I purchased my boards from them - you may need to cut one trace and add one jumper.
(8) Wakefield 423A 0.67c/w Heatsink - MPJA #12671 HS $56 (click on the "Heat Sink" link, stock varies frequently) This heatsink measures 4"w by 5.5"h by 2.65"d and has 8 fins. Total radiating surface area = 275 square inches - enough to dissipate about 25 watts of power. Virtually any heatsink will do, but keep in mind that each transistor needs a minimum of 1.0c/w. Choice of heatsinks will likely affect the final size and shape of your chassis, so I'd recommend purchasing your heatsinks before trying to design your chassis layout.) There are a number of links at the bottom of this page to surplus vendors, many of which carry a random assortment of heatsinks.
(4: Ca-d) 24,000uF 50v Computer Grade Caps - MECI $24 Click on "Capacitors" then on "Computer Grade" stock varies frequently - you might also want to check our Apex Jr. for caps - don't forget to order some mounting brackets for your new caps! (see note about surplus caps in construction notes, you may be better off with new caps...)
(8) Bergquist K-10 TO-3 Insulating Sil-Pads 0.20c/w - Digikey $20 Search on "Bergquist" or on "Sil Pad" (with Sil-Pads, you don't have to mess with heat sink grease...)
(2) Bergquist K-10 Rectifier Insulating Sil-Pads 0.20c/w - Digikey $3
(8) Keystone 4601 Chassis Mount TO-3 Transistor Sockets - Digikey $12 Search on "Transistor Socket"
Subtotal: $37
(2: Q11) 2N5248 FET - Ack Electronics $2 see note about pin out for this FET below
(10: Q1-4,6) MPSL01 NPN - Ack Electronics $8
(2: Q5) MPSL51 PNP - Ack Electronics $2
Subtotal: $12 (use Ack's search engine for these items)
(1: F1) 5A 250V slow blow fuse - Parts Express $2
(8: F2-4) 500ma 250V fast blow fuse - Parts Express $4
(1) RCA Chassis Jack pair- Parts Express $6
(2) Speaker Binding Posts, Insulated - Parts Express $10
(10) 1/4" by 1-1/4" Fuse Holder - Parts Express $11 (search on P/N)
Subtotal: $35
(4: C2,3) 220uF, 10v Tantalum Capacitors - Newark Electronics - $65 (see upgrade note below)
(4: Q9, 10) MJ11015 PNP Transistors - Newark Electronics $15
(4: Q7, 8) MJ11016 NPN Transistors - Newark Electronics $15
(1: S1) Corcom 10CFE1 10A EIC Power Entry/RFI Filter - Newark Electronics $25
(2) Metal Oxide Varistor AC protection - Newark Electronics $2
(1) Six foot EIC Power Cord 14ga #37F3337 - Newark Electronics $6 (this is a very heavy power cord)
Subtotal: $134
(2: R20) 10 Ohm 1W 5% Carbon Resistors - Web Tronics ($7 Minimum) RG10 $1
(8: R16-19) 0.68 ohm 5w 5% Power Resistors - Webtronics PW5-.68 $3 (see upgrade note below)
(32: R1-15, R22) Metal Film Resistors 1%, 1/4 Watt - Web Tronics RC-xxx $5 (see upgrade note below)
(2: C4) 39pF 5% 300v Silver Mica Capacitors - Web Tronics DM10-390J $1
(2: C1) 300pF 5% 300v Silver Mica Capacitors - Web Tronics DM10-301J $1
(2: C5) 0.1uF 100v Mylar Capacitors - Web Tronics 23BK410 $1
(2: Ce,f) 0.01uF 500v Ceramic Capacitors - Web Tronics 21FA010 $1
(2: Da,b) 35A 600v Bridge Rectifiers - Web Tronics MB356 $5
(6: D1-3) 1N4148 Diodes - Web Tronics 1N4148 $1
(2) 10k 1/4W Miniature Vertical Mount Trimmer - Web Tronics 32VQ401 $1 (not needed for stock bias, but useful for different bias points then replace with resistor after setting)
(8) 150pF 50v Rail Decoupling Capacitors - Web Tronics 21CB150 $2
(8) 0.01uF 500v Ceramic Disc Snubber Capacitors for Rectifiers - Web Tronics 21FA010 $2
(4) 6v 5% 1/2w Zener Diodes (input static protection) - Webtronics N5233B $1
(1) Soldering Heat Sink - WebTronics $2 (a self-clamping pair of tweezers also serves the same function, either way, a very highly recommended soldering tool!)
Subtotal: $25
Approximate total parts cost: $600 (US)
So this takes us to an approximate (the cost of smaller items has been rounded up) total cost of $600 (USD). Using this list, it will be necessary to order parts from a total of 10 vendors, so figure another $50-60 for overall shipping, bringing the grand total somewhere near $650. You may be able to further consolidate the vendor list and therefore reduce shipping costs.
Notes About the Parts List:
Upgrading Parts: The use of Tantalum capacitors in amplifiers has drawn some discussion and controversy. Nelson Pass has indicated that they are among the worst and actually measure poorly. Some have suggested that replacing these with equivalently rated Black Gate caps or other high-end capacitor will yield sonic improvements as well as reduce the cost. For example, if you wish to use more exotic resistors and capacitors for this amplifier, you might want to have a look at Black Gate polarized caps instead of the tantalums, Caddock, or Dale precision film resistors instead of standard metal film resistors, and MILLS non-inductive wire wound resistors. Check the link for Michael Percy below for these parts. Note that these substitutions will significantly increase the cost of the amplifier.
Items not on the above list include: (optional) power-on LED indicators, shielded wire for the input signal, and heavy gauge wire for the output signal. That's it, EVERYTHING else is listed above!
The power supply parts have been over-rated (for a very minimal increase in price) for a little extra reliability given the original design. Examples include 65v filter caps, 35A 600v bridges, and a 600VA transformer. If you use toroids, you'll need to use two stock transformers (one for each channel, 300VA minimum - 400VA might be better if you plan to increase the bias level) to arrange the center-tap power supply. An alternative is to use a custom-wound transformer with two center-tapped secondaries (4 secondiares wired in series two at a time). One is available from Toroid.com, it is Transformer #6688 and contains one 120v 6.2A primary, and two 22v, 16A center-tapped secondaries (a custom order from Nick Sirkin). Nick comments that this transformer works fine. The cost of the custom toriod works out to be about the same as two smaller toroids, so the only real issue is space in the chassis for 1 vs. 2 transformers. Also, for an extra charge of approximately $5-10 each, a magnetic shield can be added to the transformers to further reduce electromagnetic radiation.
Finally a word of caution with the power supply: Be careful about over-rating the voltage of the power supply as you don't want to go beyond the maximum voltage ratings of the small signal transistors, especially the 2N5248 FET which has a maximum voltage rating of 30V. See the Data Sheets below for details. Additionally, several people including Nelson Pass have indicated that the output of the BJTs are most linear when their quiescent current is less than 1 amp. Nelson also comments, however, that although the amplifier is less linear at bias levels above 1A per transistor, it often sounds better... Back to the tradeoff of sound vs. measurement...
Also, I've added a few things not found in the original plans: 1) snubbing capacitors to help reduce noise from the bridge rectifiers, 2) back to back zener diodes to protect the input from static zaps, and 3) rail decoupling capacitors to help with RFI rejection. Thanks to Robert Pear & Mark Finnis (from DIYaudio) for the suggestion for input diode protection, JonT (from DIYaudio) for help with heatsink calculations, and the advice of many others along the way!
All of the above parts reflect Nelson Pass's original parts list, with the single exception of the Motorola output transistors (MJ11015 and MJ11016), though Nelson does recommend these as one possible alternative. Using the Motorola (now On-Semi) devices results in a bias level slightly below what is indicated in the article since the originally specified Lambdas have internal resistance of 25 ohms, whereas the Motorolas are 40 ohms. If you feel the need to adjust the bias point, Nelson recommends a slight increase of R11 or decrease of R12 to increase the bias, and vice versa.
Bias and Thermal Issues:
One way to to determine proper bias for various output transistors, you can temporarily install a 10k pot in place of R11 (4k75 ohms) in order to allow easier adjustment of the bias current. Once properly adjusted, replace the pot with the appropriate value resistor. Using your amp meter, simply place it in series with the output transistors that draw from the power supply caps (the target draw is approximately 0.8A per transistor). Alternatively, you can determine the bias setting by measuring the dV across any of R16-19. Using Ohm's law:
I = V/R = Measured dV/0.68 ohms
Projected bias is about 1.6A per output pair (from the article), or 0.8A for each transistor each, so
dV = ~ 0.8A x 0.68ohms or approximately 0.55V for each of R16-19
Using the specs provided in the article, the bias will be properly set when the voltage drop across any R16-R19 measures about 0.55 volts. If your measurement is different, adjust R11 or R12 as described above. Just be sure to allow approximately 1 hour for the amp to warm up before measuring the initial bias. Then, make adjustments slowly and allow about 15 minutes between each small adjustment before measuring the bias again. The table below indicates several measurements of the bias setting using the above formula, my volt meter, and a temperature probe. For all of these measurements, the AC voltage into the transformer is 121.1V, producing 24.2V on the secondaries of the transformer, resulting in 31.2V rails.
Thus, using the value of R11 as specified in the article (4k75) along with the Motorola Output Transistors (40 ohms internal resistance) yields a bias level that is a little bit low (see table above). Depending upon the size of your transformer and heatsinks, slight increments in the bias level may be effected by substituting various higher value resistors for R11. Please note that to increase bias levels above what I have presented here really requires a transformer larger than 300VA per channel. I finally settled on a value of 5k6 for R11, providing a bias level of about 3.1A per channel.
After about an hour of warm up, the top of the output transistor cases run at about 50c, the top of the heatsinks run at about 46c, the transformer itself runs at about 36c, the outer casing on the rectifiers run at about 49c, the power supply caps are at 29c and my home made heatsinks on the small signal transistors mounted to the PCB run at about 40c. All of these measurements were made with an ambient temperature of about 22c and are well within the safe operating area of the respective devices. Current draw for the entire amp is 1.57A from the wall, and each pair of output transistors in series draws 1.57A.
Some Useful Articles:
Electrical Grounding 1
Building Power Supplies 1, 2, and 3
Heat sink selection & transistor mounting
Surplus Parts Vendors:
MPJA, C&H Sales, All Electronics, MECI, Surplus Sales of Nebraska, Skycraft, MCM Electronics, B.G. Micro, Fair Radio Sales, Excess Solutions, Electronic Goldmine, Alltronics, Jameco, HSC Electronic Supply, Brigar Electronics, Apex Jr., Cascade Surplus, Excess Solutions, Darrah Electric
General Retail Vendors:
Digikey, Mouser, Newark, Allied,
Boutique Parts Vendors (exotic resistors, capacitors, etc.):
Michael Percy Audio, Borbely Audio, Angela Instruments, Audio Note, The Parts Connection, Welborne Labs, Williams Hart Electronics, Sonic Craft,
Report abuse