pac1085's rambling test equipment thread

[pac1085]

Ok, not really a repair but I bought a bunch of chokes for my Dynaco's. I have two Triad C-24X, two Hammond 156R and two Dynakit Parts C-354. I was trying to figure out how to measure them on any of my LCR meters, just out of curiosity. Actually more curious about measuring the crusty old original Dynaco chokes. Even with the LCR meters at 100 Hz you can't measure them(They show like 5H) since you need to apply a DC bias to them to get them near their rated current (200-250mA).

My 4192A has an internal bias supply up to 35v but can't provide enough current. The 4274A has a connector for external bias supply but only allows up to 100mA.

I found this old HP application note about providing external bias for testing inductors up to 10A. It's pretty involved though, so not sure I'm going to bother.

https://hpmemoryproject.org/an/pdf/an_346.pdf

Looks like there are easier ways, but it would be cool to use the LCR meters.
I also found this video:

And document: https://dalmura.com.au/static/Choke measurement.pdf

 

[GKTAUDIO]

Hey, Phil! Just saw this thread and will check out your YT channel.
Greg

 

[pac1085]

I ended up building 2 of the 3325 option 002 high voltage output boards. They are essentially a 4x gain amplifier. 3325 can output 10vpp on its own, this bumps that up to 40vpp, but specified as >500 ohm load.

I wasn’t able to test them in my 3325 since I’m missing the power cord. I read online that it is installed in all of the 3325s and just tucked away if they don’t have this option but that wasn’t the case. Looks like I have to buy some more molex connectors and pins on my next order.

Also, wish i splurged on the gold ENIG plating on the boards so they looked like original Hp, considering the person who designed them left the solder mask off the top for that reason

 

[pac1085]

Well, I can't believe I actually did it, but I fixed the Keithley 6517A Electrometer!! (the one that Tektronix claimed was unrepairable)

I took my chances and spent about $75 on 4 parts at digikey. Once I got the parts and took the board out, I could see that someone had already removed/reinstalled the LMC6001 chip and LS313 dual npn transistor package. They did not clean the flux or anything...


Anyhow, I removed the LMC6001, the LS313, the two LT1013's. I then thoroughly cleaned the board and installed the new chips.

I noticed that the old LS313 (it's being used as some sort of dual zener clamp, the collector legs are cut off) - was not matched between sections. The DCA75 showed it as zener diode like I expected, but 5.99v @ 5mA on one half, and 7.588v @ 2mA on the other. Where as the new LS313 measured exactly 7.027@2mA/7.028@2mA. I'm guessing that was the issue. I'll never know about the other LMC6001 as I have no way to test it. The LT1013's were just for good measure.

I also replaced two caps on the +5 and -5v rail, 100uf 25v. They were reading a bit low at 70uf and the ESR a lot higher than the Panasonics I put in. I did that because when I measured the power supply earlier on, the ripple on those rails didn't match each other where as others in the system did.

Anyways, it now passes offset-adj and the readings are stable! I'm thrilled. Would it pass a full cal? Who knows. I hear these things are nightmare, any stray dust, flux residue, oil from your hands, etc can throw it way off. I cleaned the board the best I could with Methanol at the end.

All the work I did is in the top right corner, by the metal can LS313 transistor.



Bottom of the board after cleaning. It could have used a little more but I got it really nice in the most sensitive area.



I might see how much it costs to get this calibrated. It might be worth it, considering its value.

 

[Cubdriver]

Crazy that anyone with even a lick of knowledge about test equipment could have thought leaving that ugly mess of flux on the circuit board of an ELECTROMETER was in any way, shape or form acceptable. WTF? Did whoever did it figure they cut those slots in the board material to save weight? Sheesh!!! Nice fix, congrats, Phil!

And you've inspired me to move the 428B clip-on ammeter up in the repair queue - just went down and looked at it - it was last calibrated 40 years ago, in April of 1986. It should be fine, right?

-Pat

 

[GKTAUDIO]

It is amazing, and irritating. I've seen this with a few of my pieces of test gear. Maybe the person figures it just has to work, doesn't need to pass cal? Dunno. Lots of good gear has been decommissioned and is picked up relatively cheaply at auction, or just sits on shelves until someone tries to "fix it" to put it back into service. I've seen really shoddy work done to analog oscilloscopes, power supplies, and expensive test instruments like my HP 4274A - some of it as poorly executed as audio repair by uncaring "techs".

I'm currently trying to repair a LeCroy 44Xi 400mHz scope that had some careless damage to the rotary encoders, then further damage by some hammer-jockey who broke pieces off of the top corners of the front panel. I've got one encoder replaced, but can't find one for the other that had the shaft completely snapped off inside the housing.

Phil - you are getting to be quite the whiz at test instrument repairs! I've repaired my 4274A but had a donor unit to help diagnostics by swapping boards. Repaired my Tek 577 and use it regularly, but that was pretty much power supply and switch related. I'm not up to speed on diagnosing these units with a bunch of chips in them, but whatever I do work on is left immaculate - no flux, skin oils, etc. My stock of non-op analog scopes is frustrating, but my two big, fast expensive Tek digital scopes (1mHz and 3.5mHz) have me stuck and from what I've read on eevblog, may be hugely expensive to fix.

Keep up the interesting work, Phil!

 

[pac1085]

Thanks guys, I'm absolutely thrilled that I got the thing working!!

Yea... I'm not sure about the Electrometers previous repair. As far as I know the last place to touch it was Tektronix themselves. It came with the service receipt when I got it. No date on it, though. I talked to someone on a Discord server that has a friend who works for Tek, he said they would only replace the entire analog board, not do component level repairs, so not sure. Maybe the owner tried working on it after they got it back from Tek. Although not sure why they would have desoldered/resoldered the bad components in.. shrug. And the company that owned it certainly has enough money to just replace it.




I was playing around with it some more the other night. Measured input impedance of my HP 410b. I expected it to measure a little high because I know the 22M carbon resistor in the DC probe has drifted high. I haven't replaced it yet, because the meter calibrates fine. I have one here for it, but not sure if using a metal film in that spot its the best choice? Ignore the reading on the 410B, I didn't zero it since I didn't care for this test.



On a more audio-related note, I also tried measuring the reverse leakage of some 1N4007's. I have these old ones with gold plated leads that came with some stuff I bought a while ago. Many of them do not check out, they are leakier than the spec of 5uA @ 1000V.... many of them closer to 20uA and continue to climb.. Where as a brand new production 1n4007 was only measuring ~52nA! I need to build an enclosure with an interlock since these tests are kinda sketchy at 1000v


I ordered a new (well, used good - hopefully) VFD for it. I had to buy an entire Keithley 7001 switch unit to scrap it out of. It uses the same VFD as the 2001, 2002, and a few other Keithleys that use the dot matrix display. The seller tested it for me and sent me this. I think it's pretty good. I just hope his camera didn't over expose it. He said it's bright, though.


As compared to mine on the same test (looks worse in person):




On another note, today was the first hamfest of the season and look what I got @Cubdriver
Go figure I paid half as much for this whole thing than I did for the one probe last week...



As if I needed another one This ones slightly newer, I see 1980 date codes on the capacitors. I think the probe may be a little damaged though, there appears to be a chip in it. It does power on and kinda work but something is off with it. I plugged this probe into my good unit and it does work, although it's a little strange on the lowest range. We'll see!




Haven't installed the HV board in the 3325B yet, but I did find that I needed to swap two of the 10k resistors out for 0.1% <25ppm parts. Luckily I have a drawer full of Vishay PTF56 10k resistors @ 5ppm. I have a bunch of the fancier foil ones too but they are overkill, I will save them for when I build a voltage reference or something (there was a bunch of LTZ1000's in that same parts bin!)

 

[Cubdriver]

You inspired me to pull mine out and put it on the bench. I see yours has one of them fancy-schmantzy pee-cee boards! I think mine if from the first year of production for the breed; the serial number is 131-00443 and the manuals appear to start at the 131 prefix, so it might be the 343rd to come out of the factory. I see the 428A in the 1960 catalog (didn't try to dig back further) and the 428B first appears in 1961. Its list price back then was $550. I've ordered a POH for it from the Bay of Malcontent.

Nekkid oblique view:

I powered it up on the variac and it came to life, though it's reading low. I expect I'll be replacing some electrolytic caps, thankfully there are only about nine of them including the main filter cans, so it shouldn't be too much of an undertaking. They look a bit past their prime:

Other than the furry electrolytics it appears to be in pretty good nick. Still need to determine if any of the tubular caps are leaky, but I think these are probably film ones:

Think I might scare up some three conductor plus shield at some point and rewire the probe. It's not brittle and break-y yet, but it is definitely not as lithe and supple as I suspect it was six decades ago. I do like the addition of the pieces of spiral wrap at the ends, further acting as a strain relief - clever:

It'd been sitting downstairs on the shelf forever which is silly - I think it will be a VERY helpful troubleshooting aid! Thank you for the prod to action!

-Pat

 

[pac1085]

So, the Keithley 7001 had a junk display, worse than mine - they are hard to photograph and overexpose easily. It looked nothing like the picture. The seller sent me another 7001 but its screen was even worse than the first one.

A totally awesome member of the EEVBlog forum came to the rescue and hooked me up with a nearly brand new VFD! He didn't know if it worked or not since one of the transistors on the PCB was burnt up, but I desoldered it and swapped it into my unit and its PERFECT!!




Also, I don't feel so bad about wasting money on the 7001 switch unit(s) now as I found a use for them. I found 4 brand new 7054 high voltage scanner cards super cheap. I think it's one of the only high voltage scanner cards out there, 10 channels @ 1000Vpk. Most other scanner cards are limited to 250 ish volts it seems. I'm happy!



I played around with the 7CT1N curve tracer plugin that I posted about in the latest test equipment acquisition thread. It worked great but I don't need it since I already have a bunch of curve tracers. I sold it to a very "curious" YouTuber so perhaps you'll see it featured again ;-)



Picked up this Stanford Research SR630 16 channel thermocouple monitor. It actually functions as a dc voltmeter as well for up to 100 VDC. Isolated differential inputs. Will be cool to use as part of my automated test rig that I'm working on - people always seem to bring up temperature when doing comparisons of tube amp components so I'll log the temp at various places on the amps while recording other data.

It's labeled "SensArray Corporation" instead of Stanford research so it flew way under the radar price wise. It looks brand new and works great too, I'm really happy.
I have a spool of Type T thermocouple wire en-route... will have to get out to the garage and TIG weld some junctions soon.




The 3577A VNA works great too although I don't have much of a use for it currently. Hmm.

 

[ConradH]

I love those 428 DC current meters. I wrecked my first one when the probe slipped down an insulated wire to the exposed HV on a tube amp. I bought another one from the late Don Lancaster, that I still have. A note on that electrometer. It looks like that polystyrene cap is a bit crazed from solvent. There's a good chance it will develop high losses or fail in some other way at some point, so keep an eye on it.

 

[Cubdriver]

I played around with the 7CT1N curve tracer plugin that I posted about in the latest test equipment acquisition thread. It worked great but I don't need it since I already have a bunch of curve tracers. I sold it to a very "curious" YouTuber so perhaps you'll see it featured again ;-)

View attachment 3757620




The 3577A VNA works great too although I don't have much of a use for it currently. Hmm.
View attachment 3757612

Pretty blue phosphor!!!

Re the 3577A - it is better to have it and not need it than to need it and not have it! (The pack rat's mantra and the reason I'm at least a bit concerned for the structural integrity of my house...)

Oh, and having it is an excuse to look for a 35677A S-parameter test set.

-Pat

 

[pac1085]

I built a break out box for the Keithley 7001 with one 7054 card installed. It came out OK.

I also got all of the python code figured out to trigger the scan and pull the results. I am using the "trigger link" cable between the two units in addition to GPIB, so it operates much faster/more consistently than using GPIB only.

I also got the scan figured out for the SR630 thermocouple scanner. This one was a bit of a pain since it doesn't respond properly to many GPIB Commands. I had to add some manual checks to the code to make sure it actually switched to the channel I wanted the data from prior to querying it. You would type in "MEAS 1" and its would return the temperature for channel 2... was strange. It's good now though.


I am doing all this as part of my automated test system for tube rectifier performance.. lol.. long story, but should be pretty neat when it's done.

 

[pac1085]

btw for anyone interested here is the rough scope of what I'm doing with automated test equipment/measuring rectifier performance in this Dynaco ST70 test bed:
it all started with this thread: Max power test - JJ GZ34 vs Eico Gt Britain (Mullard?) GZ34 vs Solid State Rectifier Distortion performance in the same Dynaco ST-70

5AR4 Vacuum Tube Rectifier Evaluation Script — Rev 6
Instrument connections per rectifier_scan.xlsx.
K7001/K2001 scanning via trigger link cable.
HP 3562A dynamic signal analyzer added for speaker-terminal hum analysis.

═══════════════════════════════════════════════════════════════════════════════
INSTRUMENT / CONNECTION PLAN
═══════════════════════════════════════════════════════════════════════════════

HP 6813A AC Power Source / Analyzer (GPIB)
- Supplies 120 VAC / 60 Hz to the ST-70
- Measures: line Vrms, power factor (logged, not scanned)

Keithley 7001 Switch Mainframe (GPIB)
- Card slot 1: 7054 10-channel card
- Channels are closed one at a time; K2001 reads through the closed channel

Keithley 2001 7.5-digit DMM (GPIB) — reads all scanner channels
Ch 1 : VAC — Power transformer secondary pin 4 to ground
Ch 2 : VAC — Power transformer secondary pin 6 to ground
Ch 3 : VDC — Left OT primary current (HP 428B output, 3 V range → mA)
Ch 4 : VDC — Right OT primary current (HP 428B output, 3 V range → mA)
Ch 5 : VDC — Bias voltage, left channel
Ch 6 : VDC — Bias voltage, right channel
Ch 7 : VDC — B+ at 30 µF cap (first filter, from rectifier) + VAC ripple
Ch 8 : VDC — B+ after choke (20 µF) + VAC ripple
Ch 9 : VDC — B+ "A" rail to 7199 pentodes (20 µF) + VAC ripple
Ch 10: VDC — B+ "B" rail to 7199 triodes (20 µF) + VAC ripple

Note: Channels 7–10 are read twice — once in VDC mode, once in VAC mode —
to capture both the DC level and the superimposed ripple voltage.

HP 428B Clip-On DC Milliammeter
- Two units: one per output transformer primary center tap
- Output: scaled voltage on 3 V range (read via K7001/K2001 channels 3 & 4)
- Calibration factor set in HP428B_SCALE_MA_PER_V below

Stanford Research SR630 Thermocouple Scanner (GPIB)
- All channels: Type T thermocouples
Ch 1 : Ambient temperature
Ch 2 : Left output transformer temperature
Ch 3 : Power transformer temperature
Ch 4 : Right output transformer temperature

Tektronix TDS5054B Oscilloscope (GPIB)
CH1 : Tek P5210 differential probe
→ Rectifier pin 4 to pin 8 (anode A to cathode)
→ Forward voltage drop, half A; conduction angle
CH2 : Tek P5205 differential probe
→ Rectifier pin 6 to pin 8 (anode B to cathode)
→ Forward voltage drop, half B; conduction angle
CH3 : Tek TCP202 current probe (TekProbe BNC)
→ Output of rectifier pin 8 (DC load current)

HP 3562A Dynamic Signal Analyzer (GPIB address 5)
CH A : Left speaker terminal (across dummy load, direct connection)
CH B : Right speaker terminal (across dummy load, direct connection)
- Span : 0–200 Hz, 400 lines (0.5 Hz/line resolution)
- Window: Flat-top (±0.01 dB amplitude accuracy at 120 Hz)
- Mode : dual-channel FFT power spectrum
- Reports: amplitude at 120 Hz (dBV and Vrms), full spectrum saved,
L/R balance at 120 Hz, noise floor at 120 Hz bin

Connection note: dummy loads (8 Ω / ≥ 25 W) must be connected across
both speaker terminal pairs before powering the amplifier. The 3562A
connects directly across the dummy load terminals. Its input is rated
±25 V peak; auto-ranging is enabled to protect the front end.

═══════════════════════════════════════════════════════════════════════════════
MEASUREMENT STRATEGY
═══════════════════════════════════════════════════════════════════════════════

For each tube under test:

1. Warm up (600 s) with all output tubes installed at nominal bias.

2. No-load pass: bias at minimum.
- All 10 K2001 scanner channels (DC + ripple where applicable)
- SR630 temperatures (all 4 channels)
- Scope: forward drop waveforms (CH1, CH2), DC current (CH3)
- 3562A: 0–200 Hz FFT spectrum, CH A & CH B; extract 120 Hz amplitude
- 6813A: Vrms, power factor

3. Idle-load pass: output tubes at nominal idle bias, no signal.
- Same measurements as above.

4. Calculate:
- Ri = ΔV_B+ / ΔI_load (V from Ch7 DC, I from TCP202 on CH3)
- Forward voltage drop per half-wave (mean of CH1, CH2 measurements)
- Conduction angle per half-wave (duty-cycle measurement on CH1, CH2)
- OT primary currents (mA from HP428B via channels 3 & 4)
- Ripple attenuation through the LC filter (Ch7 ripple vs Ch8 ripple)
- 120 Hz hum at speaker terminals (dBV), L/R balance

5. Optional --conduction: narrow-timebase conduction angle analysis.

6. Log everything to CSV; save scope screenshots and 3562A spectra per
condition.

 

[pac1085]

I picked up this nice Tek 2467B a few years ago. It's a later SN model with the SMD A5 board. The previous owner attempted to repair it but the capacitor leakage had damaged it beyond repair.

I bought another A5 board from an online refurbisher, but unfortunately did something really stupid and accidentally wiped the Dallas ROM chip with the calibration while attempting to replace it. (Long story, but I had a defective GQ-4x4 programmer). I ended up buying an Xgecu T48 programmer and loading someone else's 2467B rom from eevblog. I also rebuilt the entire power supply board.

The scope works now but is way out of calibration.

I've heard horror stories about calibrating these without the proper test gear but I have no choice but to try since it will cost >400 if I send it out for a cal.

I think I can do it with the following gear:
Siglent SDG1032x (for time marks etc)
Leo Bodnar LBE-1322 pulse generator
Rohde & Schwarz SML-01 (for high frequency sine waves)
Fluke 5101B calibrator (For dc voltage levels)

Fingers crossed, it would be a cool scope to add to a bench.

 

[Richard P]

Good luck with the calibration. I've always wanted to acquire the equipment to do scope cals and will probably get there some day. I had a PG506 years ago that I sold because I didn't think I'd ever use it. Wish I would have kept it. I'm partial to Tek TM500 gear so I'll probably start trolling the auction site this winter to see if I can put together a basic calibration rack. My Tek TDS380 currently works well, but the Dallas IC is probably not long for this world so when that goes, I'll definitely want to do a calibration on it after installing a new NVRAM. I do have an Xgecu T48 programmer to read the current contents as a backup plan but doing a calibration would be my preference . . . keep us updated on your progress.

 

[pac1085]

Thanks I'll let you know. I played around a bit tonight and I'm not off to a good start. I think I have the equipment needed to do it but I'm having a hard time understanding what some of the instructions are asking me to do / what the proper result should look like on the screen.

I've so far been looking at the Military calibration manual TB 9-6625-2295-35 since I figured it might be easier to digest than the original Tek manual, but I think I'm going to have to revisit the original Tek manual.

I wish someone had a video showing the procedure on Youtube, or a website with pics, I can't find anything for the 2445/2465/2467's.

 

[arts]

Thanks I'll let you know. I played around a bit tonight and I'm not off to a good start. I think I have the equipment needed to do it but I'm having a hard time understanding what some of the instructions are asking me to do / what the proper result should look like on the screen.

I've so far been looking at the Military calibration manual TB 9-6625-2295-35 since I figured it might be easier to digest than the original Tek manual, but I think I'm going to have to revisit the original Tek manual.

I wish someone had a video showing the procedure on Youtube, or a website with pics, I can't find anything for the 2445/2465/2467's.

I know the feeling, my 2465A has just developed the TEST 04 fail 13 code. Not looking forward to this....

On a positive note, I just watched a demonstration video of the 2465A. He was still getting a trace at 1 gig

 

[pac1085]

I actually sorta got it. I made it thru CAL 01, 02 and 03. Now it's a lot better. No more errors at power on. Amplitude and timebase accuracy is good enough for audio work for now. After my first try I had some issues with TEST 04 FAIL 02 on power on. When that happens I guess you need to guess which cal routine you failed at. I guessed 02, since I think I made a mistake when I was supposed to switch to channel 2. I was right because it now powers on with no errors. I'm glad I didn't have to do 01 again because that one was a pain in the ass, especially the counting to 28 time marks on the screen several times over.

I was able to use my Siglent SDG1032X for most of the steps, but it can only generate pulses up to 12.5 MHz so I ran into some issues during CAL01 where it called for 10ns, 5ns and 2ns time marks. I generated 100, 200 and 500MHz sine waves from my RF generator to get thru it for now.

For the 10V amplitude steps in CAL02 I had to add my HP 467a amplifier in to the mix, I did a 2v signal from the sdg1032x and 5x amplification.

I'm feeling a lot better about it now, it was very intimidating for the first few hours. I might look for a time mark generator or a fancier AWG.

 

[Mangler]

Thats a beast of a scope glad you got it usable. Looks like its in really nice shape. I know its one scope that is hard to calibrate from what I have heard. My 2430a took like 10 minutes to calibrate after replacing the ROMs.

 

[pac1085]

I had a PG506 years ago that I sold because I didn't think I'd ever use it. Wish I would have kept it.

Yeah...I've been having this feeling a lot over the past 6 months. I sold a whole bunch of stuff for little money after convincing myself that I'd never use it, but time and time again, a few months later I find some reason why I should have kept it. Many times I end up buying it again at a higher price.

I'm just going to go full on hoarder at this point. Lol.

I have a dead SG503 hopefully I can revive that. I just bought a dead TG501 on eBay, well it powers on but doesn't produce any output. Hopefully I can fix that but I'm a little worried since the seller himself seems to be pretty technical. Between those and the Leo Bodnar pulser that I already own, I think I can get this thing pretty darn close to proper.