HP8565A Spectrum Analyzer Repair
2020/1/8
This is a monster, big, heavy and mysterious device. The dealer said that it could power up, and no more guarantee. I got good bargain then, less than 150$ include delivery. It's been about half year since i got it in 2019's summer.
I remember after power it on, a loud hum bursted. I pretty sure it's the transform. I immediately turned it off . I noticed, except several led blinked, no trace showed up. I figured the line voltage is wrong.
Line voltage selection is 120V , but i got 220V. Lucky, before i dig into how select the line voltage, a 220V to 110V transformer come to rescue me. Still, no trace, good sign is the frequency windows responded to tune knob.
Line Voltage Selection
The dealer must plugin 220V wall line and power it on for a long time, seal material flow outside from the transformer. I was afraid it had been so damaged, it might could not back to work again.
First step, select right line voltage, of course. Reference Service Manual : Line Voltage Selection, page 28
Voltage showed on Top-left is the one activate !
After choose suitable line voltage, device calmed down finally. NO trace showed up, of course. First LO output totally dead. Not so good.
Quick check
Since there is no trace, so first thing, checking the high voltage module. Test point A seems not right. Pull out the module, obviously the FUSE is good. And finally i notice the 158V voltage was gone.
Reference SM, page 440.
Open the back cover, located the main power supply board, a row of test point lined there. I found the 5V, 30V FUSE and 158V FUES blowed up. Replace with a new one, then 5V recovered, but 30V still down. 158V dead -- ~30V .
30V supply down
Supply schematic is in the last page of Service manual.
Voltage on U9 pin 7 is 45V, and pin 6 got 15V, output test point is also 15V. Pin4 of U9, is floating, other than expected 10V. Parrallen a new zender with VR15 did nothing. So, there must no current flow through the U9.
U9 : LM741, replacement might be: TL071, NE5532
I finally got a uc741, and 30V recovered. Because the 158V still down, 30V supply bring nothing back, maybe the YIG.
158V supply down
voltage on Q31/Q29 push pull point is seems correct --- 220V, obviously It's trying to drive Q9. Base of Q28 is almost zero, and easy to identified it shorted, all 3 leads, any 2 of them is shorted.
With KTC4419 soldering into Q28 position, the output still pretty low, further checking indicate the R47 is burned to open circuits. A new resistor put on it, but smoked immediately. Blow up the FUSE again.
On fixing power supply, I should've checking the Q9 first, but i did not. Now I came to Q9, found it deadly shorted. The most disturbing problem is find a suitable replacement. And here is the original transistor list and my research for alternative.
btw, Q28 is current limiting, remove Q28 could isolating current protecting circuits.
2N3738 (Q9)
2SC4544 TO-220F Ft=50 Mhz Hfe>30 100mA, 8W Vcbo 300V Vceo 300V Vebo 7V
2SC5858 1700V hFE>30 200W
2SC2688 Ft>=50Mhz Hfe>=40 Ic:200mA 10W Vcbo 300V Vceo 300V Vebo 7V
NOTE following high voltage transistor, hFE is too low, and could not supply enough current. Minimum hFE around 80mA/5mA = 16, but 30 is recommend.
2sc5387: loaded slow start from 100V, <=130V low hFE=10
2sc3320: loaded slow start from 120V , <=136V low hFE=10
KTC4419: works fine. VCBO 500 V VCEO 400 V IC 5A PC 30 W hFE(1)=20 (VCE=4V, IC=0.1A )
2N3440 (Q28)
2N3440 replacement:
best replacement MPSA42: Hfe >=40(identical with 2N3440).
2SC2688 : 0.2A, 300V , 50Mhz (EBC), hFE >=50
2SC1507 (300V 1W, BEC, hFE>50),
low hFE: MJE13002 , 2SC4913, MJE340(BCE)
2N3054/2N3055/2N6055(Not 158V)
https://en.wikipedia.org/wiki/2N3055
3054/3055 is kind of legendary transistor. And 3055 it's own had a recent replacement, MJE3055T , or TIP33 (NPN).
2N6055 DARLINGTON 100W TO3
158V supply Recovered
2sc5387 is first transistor i picked up for Q9. Unlucky, output almost zero. Q9 , Q28 is good. I suspect the output is protected. And it did so, after isolated Q30(by lift up on lead of VR9), output reach to 220V!!!
Further checking indicate the U10 pin6 always stay on 25V ( +15V UNREG is 30V), no matter output is zero or 220V. meanwhile, U10, Pin2 is higher than Pin3 --- this suggested the U10 is broken. Another UC 741 is sacrificed for protecting other circuits.
Replace UC10 solved the problem, even though output only 125V. The scan trace emerged on screen, BIG PROGRESS. And Now is time to address why voltage is low. It's simple, though, without load it's 156 Volt. So It's because of the hFE of the transistor.
I replace the 2sc5387 with 2sc3320, but both them is not function well. finally i found KTC4419 is the right transistor I'm looking for.
2sc5387: loaded slow start from 100V, <=130V low hFE=10
2sc3320: loaded slow start from 120V , <=136V low hFE=10
KTC4419: works fine. VCBO 500 V VCEO 400 V IC 5A PC 30 W hFE(1)=20 (VCE=4V, IC=0.1A )
Almost Recover
Before any further experiment, volta must be alignment. After lightly wiggle the adjustment POT on the supply board, all voltage almost precisely on the value of the Spec.
Led display all SA setup, it's cute and convenient, almost identical as digital one. And I prefer this LED version.
Each LED display the unit, gone dark. The led control board located just beneath the CRT. Remove the screw of that board, unplug the cable so you could take it out.
I found the root cause at once, one of cable connected to the LED is totally slip off from socket. This is an happy end, and pretty weird.
Next step, going to play around my new toy. For few days investigated, I notice something miss aliment and some good things.
Symptom:
No zero beat response except on very narrow span per div.
0.01 - 1.8Ghz working well aliment panel calibrige pots.
The 1.7G-4.1Ghz band reading result is 10 dB less than actual signal level : confirmed by HP432.
More higher band: unknown.
Good one:
0.01 - 1.8Ghz band amplitude pretty precisely alignment, this mean the IF amplifier , filters, YTO driver, mixer bias still stay alignment.
All bands show scan line.
All control knob is working well, except the Attenuator knob is broken, not a big deal since we had LED display.
Trying Adjust Frequency Response - the dBm Amplitude accuracy
1.7G - 4.1G band is convenient band. I had several YIG oscillator on hand, nice to have a SA to play YIG Oscillator. Since dBm reading is almost 10 dB less than actual value, I suspect something go wrong. I initially think some alignment must be done to bring back the accuracy.
There is no zero beating, that is suspecting symptom. Service Manual give a brief table, list all adjustment Pots. I notice there is YTO/YTF adjustment. Obviously, I supposed to tune YTO offset, let the zero beat response emerge on screen.
The CAL output is accuracy enough for me (-0.45 dBm) . Reference level easily align by 100 Mhz CAL source. Tune YTO offset POT A19R5, let zero beat show on screen, then tune A19R8, YTO Gain, let all CAL sources harmonic align with vertical line.
On harmonic bands, each band use one harmonic component of the YTO, so each band associate with a number N: represented the harmonic which it gonna to use.
On harmonic "N" band, a YTF switch in as band pass filter, cover 1.7Ghz - 22Ghz.
YTF must tracking the YTO to ensure correct response. I realized this after i desperately tried to adjust the frequency response(B2A,B2B, B2C) for hours.
Band 1.7-4.1 is simply controlled by YTF-OFFSET(A19R14), and YTF GAIN(A19R17). According the Service manual, you should turn on a YTO TEST switch on A19, so the YTF response will displayed on CRT, and use sweeping signal source to help tune YTF offset and YTF GAIN. I don't have a the sweeping microwave signal source at all, I realized i could use main tune knob, under help the this switch to checking the YTF response.
While you aliment YTF response, PLEASE keep the YTF peak knob stay middle point.
I did not have any microwave signal source available. But I did had several YIG oscillators. Bench Power supply had current limiting is suitable to drive the YIG tuned coil -- for safety, serial with 5-10 Ohm resistor.
Even though tune the YTF OFFSET and GAIN is easy but checking the response is pretty awful , without a sweep signal source.
Good thing happened, I finally get band 1.7-4.1 working well.
BTW: I had advanst YIG, which output 40mW form 2G to 4Ghz, pretty powerful, and almost flat(several dents reach to 30mW). And it's harmonics also strong and extended to 10Ghz. This help me a lot a while late.
Now, band 1.7-4.1 works, I wonder, why not try 3.8-8.5, 5.8-12.9Ghz. It seems pretty simple: each band had its own OFFSET, but share same Gain, I suppose the Gain control somehow the END of one band. I reallize this is not how the tracking works, after spend hours and hours trying to adjust each offset.
Adjustment Steps
Before going any further, I better briefly reading the alignment procedure. First of all, I need figure out why a SA read correct dBm on CRT, what ensure this accuracy.
For the harmonic N band, following essential setup ensure the frequency response:
YTF must tracking YTO: This ensure a correct response
Each band had to have it's own frequency response compensation. Mixer it's self response pretty flat, this correction is very small - few db maximum.
@Page 158
Adjust should following this table. The most important group is A, B and C is some kind of linear adjustment. I guess they are not beed messed up, even they are not perfect, still could get meaningful reading after group A is finished successfully.
I spends 8 hours to figout how these things work together. Here is bold NOTES:
All band share a same OFFSET actually.
N2-N5 is just very slight adjustment: N2: 5db Range, N3: 10 dB N4: 20dB. Don't evng thing use N2-N5 aliment each individual band.
All Band share same Gain: group A's GAIN adjust should be gradually reduce to get proper response if everything is going right.
After spend desperating 8 hours, not all of them is meaningless, I gain a method could quickly check the response without a sweeper, and don't need to use TEST mode. The key idea is, the "Preselector peak" knob help you sweeping the YTF.
Quick Checking YTO response
Use YIG and it's harmonic.
YIG frequency at beginning of the band, around 2Ghz, it's harmonics is 4G, well done.
Rotating "preselector peaking" knob, if either side the response leftover is on same level, the tracking is correction.
Now, the tracking response on both end simultaneously present.
Suggest:
These group of Pots is protected by a aluminum shield on purpose.
This suggest don't' touch them, unless you know what you are doing.
While you need from these pots to get accuracy response, the error must be very small maybe less than +- 3 db.
In other words, these make the response perfect, not intend to make them just work: please make sure YTF tracking is well alignment.
How YTF tracking works
I spends hours to adjust the Pots, but could not make all band tracking correctly. I then suspect how exactly the tracking work?
First of all, they share same YTF GAIN, what's is GAIN exactly?
They then share same offset (YTF OFFSET), N2-N5 just could adjust a little bit, they are not intend to make tracking work, they intend to make tracking perfect.
I do not how to discover this, but they actually working prety simple: YTO and YTF tune line should be overlap.
YTO and YTF are pretty linear, plus it had linearity correction circuits. They are suppose 2 lines on mv-Mhz coordinate system.
YTF share same SCAN Ramp voltage which control the YTO, now the YTF GAIN jump in: YTF GAIN control change this Ramp's Slope to YTF's slope. So, YFT change same amount of Mhz with YTO.
YTF also share same CENTER voltage with YTO. YTF OFFSET gonna make YTF tune line overlap with YTO tune line: 2 tune lines already share same slope . And now , with YTF OFFSET correctly, there share a same mv-Mhz point. then, they are same line, so, perfect tracking gonna happen.
Good Practice on Adjust YTF Tracking
Knowing how does help you adjust the YTF Tracking, but not so much. I spend hours again, on how to make them alignment. I finally found the YTO OFFSET impact the tracking a lot.
And, Of Course, It does : YTO and YFO both are not perfect linear line, especially on beginning of the ramp. This explain everything: the zero span should not on the start of Sweep, ZERO SPAN should move to a little bit right. 50Mhz is minimal request. Fiannaly, offset 70Mhz give a good YTF alignments.
Following steps is tring recovery the alignment without totally go through all alignment items. This is far from perfect, only enough for hobbyist, especially no test device available supporting the alignment procedure.
Step by Steps :
Put N2-N5 on middle position and dont' touch them on tracking alignment: they can't make TRF track happen, they make them perfect.
Adjust YTO OFFSET, locate the zero span on 50Mhz-100Mhz far away from Sweep beginning. Start from 50Mhz. Then use CAL as Combo Source, Adjust the YTO GAIN.
On band 1.7-4.1, put yig signal and harmonics like the left photo. Use the "Preselector Peak"knob to checking the band ends YTF response.
Adjust YTF OFFSET, make the low end of band 1.7-4.1 response correctly.
Adjust YTF Gain, make the low end of band 1.7-4.1 response correctly.
Switch to other Bands, use YIG and "peak" knob to check band ends YTF response.
The YTF GAIN(high end response) might need be fine tuned tiny little bit.
Checking the left end YTF response, If response almost good, ie: while "peak" knob stay middle. we should got maximum response level . IF so, go next band. (On band N4, could tolerate a little bit more than this, YTF OFFSEt N4 does could bring +- 10dB back.) If checking failure, go back to step 2, put YTO OFFSET in another position, start over.
OFFSET (NOT the Nx which suggested by following Table)
How HP8565 Frequency Readout DVM works
Reference
https://blog.muwave.de/2015/04/hp-8565a-spectrum-analyzer/
https://blog.muwave.de/2015/07/hp-8565a-sweep-time-selector/
hp8569a/8565-microwave-spectrum-analyzer a great find 10mhz to 22Gh
http://www.simonsdialogs.com/2015/03/hp-8565a-hp-8569b-spectrum-analyzer-a-mixed-box/
http://www.simonsdialogs.com/2015/03/hp-8569b-spectrum-analyzer-working-essentials/
http://www.simonsdialogs.com/2015/03/hp-8569b-spectrum-analyzer-almost-the-same-not-quite-identical/
http://www.simonsdialogs.com/2015/05/hp-8569b-8565a-spectrum-analyzer-curing-the-knob-disease/
http://www.simonsdialogs.com/2015/07/hp-8565a-spectrum-analyzer-a-new-crt-for-a-trusty-unit/