Electro Harmonix Memory Man

The original Memory Man was an early bucket brigade delay (BBD) effect pedal that allowed musicians to create long echo effects without using tape.  It has been claimed by Mike Matthews (founder/owner of Electro Harmonix) that the Memory Man was the first such "stompbox" type of effect for echo.  Release dates for the Memory Man and its competition seem to be impossible to track down today, so whether or not the Memory Man really was the first of its kind cannot be verified.  If it was not the first, then it was most definitely at least one of the first.

Vintage Memory Man pedals can be roughly sorted into two groups based on the type of BBD chip used.  The early pedals used Reticon SAD1024 chips.  This group may have begun in 1976, and may have been discontinued in 1978.  From 1978-1984, Panasonic MN3005 chips were used.  The vintage units include several revisions of 3 knob, 4 knob, and 5 knob models.  4 knob models were exclusively SAD1024 based, but 3 and 5 knob models were made with both types of chip.  Below you'll find information on each variation I am aware.

Modern Memory Man pedals began with the reissue of the last versions of the 5 knob (Deluxe Memory Man) and 3 knob (Stereo Memory Man).  This may have been around 1996-7.  The 5 knob was revised a few times, and the 3 knob was deleted after a few years.  Panasonic had discontinued their BBD chips by the 2000s, but Electro Hamonix stockpiled them and reissue Memory Man production was OK for a few years.  Eventually, by the late 00s, the Panasonic chips ran out.  New BBD chips became available from the Shanghai Belling company.  These are labelled BL instead of MN.  With the new chips, Electro Harmonix released the Memory Boy, the Memory Toy, and deluxe versions.

Calibration/Troubleshooting:

• I have edited the 1978 calibration procedure from Howard Davis for the Deluxe Memory Man (5 knob) and it can be found here: EH 7850 Deluxe Memory Man Calibration

• My "rough draft" (no plans to finish!) of a 7550 calibration guide..

• Layout diagram with calibration guide for the 1370B (SAD1024) by Anders Gustavsson.

• In depth calibration guide for the 1370B by Lucien Grandjean. 

Information on integrated circuits (ICs) found in the Memory Man line:

SAD1024/MN3005

Explaining the bucket bridge delay ICs that create the delay effect.

NE570/571

Explaining the "compander" IC usually found in a Memory Man.

CD4047

Explaining the clock IC in a Memory Man.

LIST OF MEMORY MAN CIRCUIT BOARDS:

Links contain schematics and some circuit analysis.  This list may grow as more designators are discovered.

The "Original" Memory Man variations (SAD1024):

EH BBD ECHO ISSUE A - Memory Man - 3 knob - Gray box, black line graphics - Boost Switch

EH 7500 - Memory Man - 3 knob - Gray box, black line graphics - Boost Switch

EH 1309A - Memory Man - 3 Knob - Gray and Black - Boost Switch

EH 1309B

Photo of a EH1307A PCB.

The 4 Knob Deluxe with Squelch variations (SAD1024):

EH 7550 - Deluxe Memory Man - 4 knob - Black and Red - Squelch

EH 1307

EH 1307A

EH 1307B - Includes layout with calibration notes

EH 1307C

The Chorus with Boost variations (MN3005):

EH 7810 - Memory Man - 3 knob - Blue? - Chorus Switch - Boost Input

EH 1338 - Memory Man - 3 knob - Silver and Black - Chorus Switch - Boost Input

EH 1338A

Photo of an EH 7811B PCB.  The "missing" IC footprint is for a 741 op amp.  This is populated on EH 7811.

The Stereo Memory Man variations (MN3005):

EH 7811 - Stereo Memory Man - 3 knob - Black and Red - Chorus Switch

EH 7811A

EH 7811B - Stereo Memory Man - 3 knob - Black and Red - Chorus Switch - LED

EH 7811C 

EC 2020 - Reissue

Photos of an EH 7850C PCB.

The 5 knob Deluxe variations (MN3005):

EH 7850 - Deluxe Memory Man - 5 knob - Black and Red - Chorus/Vibrato - 2x LED

EH 1343B

EH 7850C

EC 2000 - Reissue.  Early versions have hard wired power cable.  Later versions use 24Vdc adapter.

EC 2002 - Reissue.  "True bypass" and relay bypass versions.

Memory Man Frequency Response

Above: You're looking at a scope trace from a sweep generator passing through the delay circuit of a Deluxe Memory Man (which one? my notes don't say..) and the sweep is from 55Hz to 5.5kHz.  Top (red) trace is the input signal, and the yellow (bottom) trace is the output.

Right away you should see that the high end is heavily attenuated.  Anyone familiar with these pedals is aware of the "dark" quality of the echoes.  What this reveals that isn't quite so obvious to the ear is the little bass cut and the strong mid boost that is also present alongside the high cut.

Some Technical Info:

All Memory Man pedals are BBD (bucket bridge delay) based effects.  Any BBD effect will consist of a minimum of:

-1 or more BBD ICs.  Early Memory Man pedals use SAD1024 chips, later switching to MN3005 chips.

-A clock circuit, usually based on a IC.  A CD4047 is usually used as the Memory Man clock.

-Op amp circuits that boost the input signal and then mix the input and delayed signals.

-A controlled positive feedback path from delay output back to delay input.

Optional items typically found in a BBD effect:

-A LFO for modulating the delay clock.  This also requires some sort of circuit that allows the slowly varying voltage from the LFO to influence the timing of the delay clock.  Once this is in place, the effect can go from being "just a delay" to also having flange/chorus/vibrato ability.  The term "flanger" is typically only applied to pedals designed with very short delay chips.  The Memory Man's delay chips have too many stages to make a decent "flange" effect.

-A compander chip (either the NE570 or NE571) to reduce noise.

-Various types of input and output filters to reduce noise.

Troubleshooting Tips:

(Additional Tips)

Bucket Brigade Delay Chips

Dead chips will increase the repair cost quite a bit.  Fortunately, a dead chip is usually not the problem.  Exhaust all methods of troubleshooting before worrying that your chips are dead.  eBay is the primary source of these discontinued chips.  "Pulled"/used chips might be anywhere from $30 to $100 dollars.  Prices vary widely.  NOS/tested/guaranteed chips might be advertised for over $100.  There are other MNXXXX Panasonic chips and clones, but none of them will directly replace a SAD1024 or a MN3005, hence the hefty price tag for the old chips.  Substituting chips requires a daughterboard.  Still cheap and plentiful MN3007 chips can substitute, but only a determined DIY'er will have this option, unless some repair person out there is offering this service for less than the eBay prices of used chips.  Since SAD1024 sections are a mere 512 stages of delay, you can simply jump a dead section for a cheap repair (there are 2x 512 stage sections per SAD1024 chip).  The maximum delay time gets reduced, but the pedal will work again.  You can also jump a MN3005, but that really reduces the max delay available (halves it) since you are skipping 4096 stages in one go.

If the circuit appears to be functioning normally (clean signal comes through, you can see the 2 clock signals coming in, etc.), but you don't get any delay, then go straight for the first delay chip.  Figuring out which is first can be confusing, so just check all the inputs with the feedback control at minimum and you'll eventually figure out the first input because it will be free from any clock noise.  Now check each input and output.  When you get to a dead output, check the clock inputs to that chip.  Next check the bias voltage to the input.  Adjust as necessary (make sure it's there and that the adjustment is actually changing some voltage at the input pin).  If socketed, make sure the chip is well seated.  Take it out and put it back in case the socket and/or pins are oxidized and have bad connections.  If you have multiple chips, try swapping them and note any changes.  If you have another pedal that accepts the same chip (old MXR and ROSS delays are a great source for SAD1024 testing) try the swap there.  After all this, if there is no delay signal, the chip is most likely dead.

The SAD1024s don't always just die.  I've caught some that would actually impart gross distortion on the signal.  The signal would come out clipped no matter the bias setting.  Swapping 1024s fixed the problem.  So it is very important to assess the quality of the BBD chip's distortion as well.  If you have a THD meter, you should be able to get pretty low numbers (<1%) with good chips, but gross distortion is easy to detect with just your ears alone.

What about the BLXXXX chips in the new Electro Harmonix delays?  As far as I know, these only come in the 3200 series variety, so they will not sub for a MN3005, or SAD1024.  These are clones of the old BBD chips.  Before I had any in-depth technical knowledge of BBD chips, I felt these newer pedals sounded worse, especially the quality of the delayed signal.  When I looked into the technical reasons, the biggest clue comes from the original BBD saga of the 1970s.  It is well documented that IC manufacturers struggled with creating chips with low enough overall capacitance to not kill the high frequencies, and FET channels that wouldn't distort badly.  Both of these issues plagued BBD and CCD technology in their development.  The muddy, crunchy sound I hear from the new ICs indicates that the modern ICs are having these same problems, but who's going to complain now?  Before the companies were dealing with all kinds of customers, but now they only have to deal with the guitar effects industry... an industry where "lo-fi" and "distortion" are often good attributes of an effect!  BBD chips are one place where I really do believe the vintage ones can sound better, but it is not guaranteed.  Vintage ones can also have bad distortion performance and high noise levels.  If you are able to get multiple chips you can audition them in circuit for the best performing ones.

Bucket Brigades Need Two Clock Signals

The clock circuit (CD4047 in a Memory) should be running above the audio range and at a large amplitude, so set your scope to look at something like 5V/div and fast enough to see kHz square waves nicely.  You should get 2 clocks.  If the 4047 isn't sending 2 clocks out, then replace the 4047 or check the supporting components, like the DELAY pot and the associated timing cap.  I have encountered more than one dead 4047 chips that seemed to have failed for no reason I could detect. 

Speaking of the clock circuit... a VERY common problem with all BBD circuits is the clock circuit getting too slow and creating an audible high pitched freq. when the DELAY control is set for long echoes.  The high pitched noise is normal as far the overall BBD operation is concerned, but it is not supposed to be there, and if you measure the echo time, you will find that you are beyond the manufacturer's claims in the delay time.  Check old advertisements and/or manuals to get the spec'd max time.  Sometimes you even have to get shorter than the promised times to eliminate all noise.

Since clock speed is determined by a RC circuit which the DELAY control is an element of, it is easy to find and adjust.  Find the "C" in the circuit, usually a pretty small cap, and substitute a higher of lower value for coarse adjustments.  You can parallel caps for fine adjustments, or you can add or remove resistance for fine adjustments.

The Compander

A detailed look at the compander IC can be found here: NE570/571.  As far as basic troubleshooting goes, the first thing to figure out is simply which side is the input/compressor and which side is the output/expander.  The compander will be placed before and after the BBD chips.  Set your feedback control to 0 (off/no feedback) and observe the signals coming and going from the compander.  If you wiggle the delay control, the input and output sections will instantly reveal themselves.  The input will not respond to a delay control wiggle, but the output will.

A dead compander chip is not something I've seen often, but I can remember encountering this at least once.  In my troubleshooting the compander really just helps identify the signal flow as I'm tracing the circuit board so I don't have to keep glancing at the schematic.  (Easy identification of main BBD input and main BBD output.)  If the compander circuit appears to be doing something weird, check the external components then review the detailed info on how the IC works.

Fragile Slide Switches

All Memory Man pedals in the classic "big box" will suffer from damaged slide switches.  Those slide switches do not take well to inward pressure.  If they are only slightly broken, you can usually push them back together.  Don't go crazy though, because you can make them too tight and break them that way as well.  Fortunately, drop in replacements are still available.  Mouser 10SS002-EV

Bypass Schemes

The Memory Man line generally used the "switch the output jack" scheme favored by many '70s pedal manufacturers.  Watch out for the confusing "effect cancel" type bypass that was used in the EH 7811 boards (Stereo Memory Man with Chorus).  Either type can be converted to "true bypass" if desired. 

Reissue Memory Man pedals have a variety of bypass options.  Some use the same switching as an old EH 7850, some use a 3PDT for "true bypass," and some have a 3PDT wired SPST to control circuit board relays to achieve "true bypass."

Misc Notes