HP400H:  A great story

This project last sevaral Monthes，from Jun to Sept there are a logs for provement. but it start before Jun, and highly possible strated from April 2025.

HP400H is classic HP vacuum VTVM, from HP, capable of measuring RMS AC voltages over a range of less than 1 mV to 300V, over a frequency range of 10Hz to 4MHz. Its not a Ture RMS device, just a typical diode detector, average-responding, rms calibrated VTVM.

Overall, the condition of this 400H seems satisfactory, 1V reading is accurate from few kHz to few Mhz. However, many components show signs of corrosion, particularly the small cylindrical filter capacitors. 

Its reading some residue while there is no input. Per HP's service mannual, It is regarded as an issue only if the residual reading is greater than 2.5 divisions. I've googled this issue, there are few articles and videos introduce some possible fixes. So, i assumed i could get this fixed in same way others have done before.

Leaking bridge capactior or Diode faulty

Someone suggested that the residual reading is most likely caused by a leakage in C32 or C33. It’s a straightforward step that can be completed quickly before proceeding with a deeper analysis.

how bridge worsk

Paralleling caps with new one

So now we've confirmed the bridge caps were not bad, and bridge diodes were good too. then, waht cause the residual reading? 

The key point here is that any AC signal will produce a reading on the meter. Since we’ve proven there is no DC leakage causing the residual, the reading must be due to some AC signal.

It took me a couple of days to figure out that the power supply ripple acts just like an input signal.

Wow, this conclusion puts me in a tough spot. We kind of proved the capacitor isn’t as bad as we thought, and we’re pretty sure the DC is fine. So that means the ripple must be there—most likely from bad caps AGAIN.

I’m doing something different this time to find the bad cap. I’m going to parallel the suspect capacitor with a new one and see if it makes any difference. See all those capacitors across each tube? I added a new cap to each, not all at once, but a few stayed in place through the whole repair process

Guess what? Every capacitor I could find muse be somehow incredibly bad, even though the bridge shows they’re all perfect, I can’t believe it, Literally no difference at all! ✅

Power Rail Ripple

Time to call in my trusty Tektronix oscilloscope. Everything points to something being off —It can’t be DC leakage, there’s clearly some AC signal causing the residual reading. I have no idea where it’s coming from. And if all these AC filter caps aren’t as bad as I thought… then what the heck is that AC signal?

Let’s see if there’s any AC signal at the output of the final amplifer stage. I’m just randomly checking here, Nothing particularly special, but it would be more obvious if any signal exists. 

And I should mention that I managed to get a backup HP400D. This HP400D works much better, and its residual reading is negligible. Having this backup VTVM — even though there are some minor differences in design and build — gives me a really good reference point for what I’m trying to do. 

With this backup VTVM, I can run plenty of interesting comparisons. The key one is the idle AC output — it might explain quite a lot. Look at the photos below — if I hadn’t mentioned that the right one is the idle signal from the good unit, would you be able to tell the difference? 

It’s kind of surprising that the idle AC signal shows no visible difference, even though one unit clearly has a residual reading. I’ve played with the idle signal for weeks — trying different attenuations set, different stages take the signal from, and diffrent oscilloscope senstivity and timming setting, — all just try to tease out the real signal that causing the residual reading from that noise-like idle trace. 

Emergence of Ripple 

I began to doubt everything, including myself. Was this a sign that my conclusions and reasoning were off? A few weeks later, I finally decided to check what the loaded signal looked like. That was a brilliant move — it pulled me out of self-doubt and gave my confidence a small boost. 

This time I tapped the output from V4 while injecting a full-scale signal. The piggyback ripple showed up clearly on top of the input signal, almost like an AM-modulated signal. 

Bad VS GOOD

Here’s another good-vs-bad comparison. The average DC level looks different between the two, but neither one is actually symmetric around zero. 

It proves two things: 

1) I’ve overplayed this too much, and

2) the idle signal can’t really show the residual AC signal. 

The loaded signal exposes the residual AC, putting us in a dilemma again: if there’s residual AC, doesn’t it suggest that some capacitors gone bad? 

Wait a moment — this feels like déjà vu. We’ve proven it’s not due to DC leakage, the capacitors don’t look as bad as I thought, and there is indeed residual AC. So which one of these is off? I fully trust the oscilloscope — that’s proof; I literally saw it happen. I also believe it’s not DC, since we already see the residual AC, so DC has an “alibi.”  Yet I can still suspect the capacitors! 

I went a bit crazy, trying capacitor paralleling for hours once more, before finally telling myself, “Nope, dead end.” It’s also possible that the AC ripple comes from somewhere else instead of a bad capacitor. Other owners have mentioned this too — one of the amplifier stages could be faulty and feeding ripple back into the loop. 

The photo above shows where I disconnected the feedback loop and began a new troubleshooting journey. It’s at V2’s cathode, pin 2 — the wire running from R105 to L10 in the schematic below. 

Oscillating Amplifier

It’s pretty easy to notice that the V4 stage seems to oscillate at a frequency close to the mains line.What would you think if you saw this? My thought was that the oscillation was causing the residual reading, and the feedback loop was partly canceling it out. Still, some of it slipped through, and that’s what the meter shows as residual.  I believed this brilliant analysis for a long time. 

I was really cheered up by this new discovery and proudly shared this big step forward with everyone. But in the following weeks, my main question became: why is this stage oscillating? I don’t want to be rude, but let’s be honest — most oscillation issues are just bad capacitors. Ha, caught me again! 

I had to do something about the capacitor. See the blue cap above? It probably won’t work as always, but there’s no harm in trying. I disconnected each stage’s input and output, replaced the tubes, checked resistor values… basically, I’m running out of weapons, as usual. 

I’m not very familiar with tubes, but the logic is the same when I suspect the amplifier begins oscillating. I don’t want to go through that tiresome debug process again, so here’s the final conclusion: you need to close the input — short the stage input with a series RC. Otherwise, because of the tube’s high-impedance input, it will pick up line radiation and either amplify it into a large signal or even start oscillating. 

Then I did the proper test, and it showed there was no oscillation at all. Furthermore, ripple voltage was present at every stage. I only noticed the last stage and the one before it because the ripple had been amplified to a noticeable level. By adjusting the oscilloscope’s sensitivity and attenuation, I could see the 120 Hz ripple on every stage. 

Oh no, I give up on suspecting the caps this time! I closed the loop just as it was at the very beginning, and everything went back to square one! 

When I look back at how this whole repair started, I have to admit I followed a pretty stupid procedure. But honestly, it was great brain training — and thinking about it now really comforts me. This should’ve been the first thing to do — check the power supply. I did it, but only very briefly. For example, I checked the B+ line; it was within spec and didn’t show anything obviously wrong. 

Now let check it follow the porcedure listed in the service mannual: 

A. METER ZERO ADJUST ✅ 

B. CHECK REGULATED POWER SUPPLY

1) The B+ voltage at the output of the regulated power supply should be 250 volts, +-5 volts.✅ 

2) Vary the line voltage between 103 and 127 volts.The regulated B+ will usually change no more than 2 volts. ✔️ (None Relevent)

2.1） Ripple voltage in the regulated B+ is usually 3 millivolts or less under these same test conditions.⚠️  A little off from the spec, but the good one actually has the same ripple level. 

2.3) Excessive ac ripple may be from a defective tube in the power supply.✅ change or switch with good one, no diffrent

Capacitor C36 may be open. ✅ its perfect and also change to a new one.

TROUBLE SHOOTING CHART

Residual meter reading steady and higher than 2-1/2 divisions on 1V scale

• Trouble in power supply (✅).

• VI, V2, V3, V4, and/or V5 defective ✅ proove to be good by swiching tubes with good one

• Defective component in amplifier ✅✅ Checked resistors and capacitors with the bridge, paralleled a new capacitor. 

• NOisy R23 ✅ Replaced — no difference

• Check mechanical zero. ✅ ofcause

• Capacitor C6 is formed by wires connecting to adjacence lug terminals. if the heater leads for VI are dressed so that they pass between these terminals, hum is introduced.   ✅ move the wire around, no difference

During those few weeks, I kept going back and forth — doing voltage alignments, checking capacitors, swapping tubes, measuring ripple, verifying voltage specs, and all that kind of stuff.

One day, I realized I’d never really paid much attention to the filament voltage. I did check the DC voltage specs though, and everything looked within limits.

Flament Supply

I’ve never seriously fixed any tube stuff before, so no wonder I ignored it. Now I’m treating the filament supply as a serious thing, checking it just like I did with the B+ line. 

Since the service manual doesn’t specify a reference level, I thought it might be worth trying a capacitor-paralleling test. Bingo! Look at the result this time — the residual reading has dropped to almost within spec.  It’s worth noting that this was on August 31, 2025. 

I have to admit, there’s a different story to tell. When I wiggled the ground pin of the filament capacitor, the residual reading suddenly dropped — and it decreased even further after I added another capacitor in parallel with the old one. I must admit, I never really paid attention to the filament supply. I thought it was just about fliment and wouldn’t have much impact on residual hum. 

The wiggling did work — it must have something to do with the capacitors having degraded, and the shielding might have gotten worse 

Development of the filament supply 

Tere are some details of the filament power supply in the service mannual of HP400C. It describes several changes and improvements that HP made to enhance the overall performance. 

on 400C there are some detail info about residual reading:

Condition A: Meter indicates the same noise amplitude on all positions of the range switch. 

The trouble source is between V2 and V5 inclusive. Look for defective 6AKS tubes, noisy resistors, defective or poorly grounded electrolytics, or defective coupling capacitors. Check all solder joints and ground connections in this area. Noise from B plus and heater supply may be introduced into V2 and V3. In this case look for poor voltage regulation noisy VR tube or other power supply tube, defective electrolytics (C2l, C22) in the dc heater supply, or defective selenium rectifier. See sections on "B+ REGULATED SUPPLY" and "DC HEATER VOLTAGE SUPPLY". 

CAUTION -- When first received from the factory or when the instrument has been recalibrated, the meter pointer may not indicate zero when the instrument is turned off. After the instrument is turned on, the meter pointer may show an indication of as much as two scale divisions with the input shorted. This effect is seen primarily on the 1 volt range and is normal. It is caused by the high gain and internal noise of the amplifier. This will not affect the calibration or accuracy of the instrument. 

I learned something new,  some residual reading is acceptable, poorly ground electrolytics might be the key — just as I observed. DC heating circuits are the most reliable way to eliminate hum if we really want it to be as low as possible.  i've learned so much from The Valve Wizard.

Each Capacitor is equally Bad or GOOD?

I checked every filter capacitor on the 400H, NONE of them have a D<0.06, on 400H the filter capacitor have D in range 0.06-0.09, while D=0.1, the ESR reach to 10ohm. 3 times worth than capacitor on the good 400D. Modern high-voltage capacitors might have a dissipation factor (D) below 0.03. I have a few one with D as low as 0.01@120 Hz.

The capacitors on the 400H definitely show higher ESR now, that's what I can tell for sure.

This is a hard way to learn what a bad capacitor look like. This exlains a lot. I've tried adding a good capacitor in parallel with the old one, but I’ve never done that for every one at same time. There are 3 big aluminum capactor cans, each one contains 2 or 3 capactors inside.

The final closure

I’ve decided to bring closure to this long-term fix. It’s actually more like a study project than servicing. I’ve become familiar with the HP400 VTVM andI have also gotten into the Tube world.

The heater supply filter capacitors contribute the major portion of the residual reading. Look at the photo above — one of the capacitors has only about half the capacitance it should have.

The accracy of the mini bridge is good enough for serviing poupours, Check this out, Below is the result from Hantek. 

Check out the photo below: after changing the heater supply filter caps, the residual reading get back within 2 divs. It’s possible to get a better result if I replace all the big can capacitors with modern ones. But one and a half divs is already a well‑accepted standard, even back when this puppy first came out of the HP store. I’ve been told that, but I’d like to get more out of it and have some fun with it. 

Summary

Identify the supply hum

• For a VTVM, connect a 120 Hz signal to the input. If you tune the frequency around 120 Hz, the signal will beat against the supply hum and make the analog meter needle wobble. It’s pretty sensitive to residual hum, and it’s a visible sign don't need a scope.

• Connect a high‑frequency signal, say 10 kHz, and use a scope to check the final amplifier’s output, just before the detector. If there’s any residual hum, it’ll show up as AM modulation

• If the hum is really high, a multimeter could show a high RMS reading on the DC rail 

Progress Recap

A simple issue

1. At the very beginning, I suspected that one or more capacitors might be faulty — purely based on intuition  

2. I’ve Googled the common causes of residual readings, and the most frequent explanation is that the bridge capacitor may be leaking. I replaced both of them, but there was no improvement at all.

3. Using a portable bridge, I measured the ESR/D values of the capacitors, one or few capacitor must have gone incredibly wrong, but they were all nearly identical, with no single unit appearing significantly worse than the rest.

4. To see why the circuit showed some residual reading, I went through the bridge design and figured out why DC doesn’t matter unless the capacitor is leaking, and how AC ends up being shown on the meter 

5. Comparing the B+ test points with a known-good HP400, there are no obvious fault symptoms when no signal is applied. still highly suspecting one of filter capacitor is faulty, I randomly paralleled good capacitor with the old one, no improvement at all evenghough i did that for quite few of those old capacitors.

6. With signal appiled, i identified the bad one have a visiable am modulating waveform, i was quite sure the power supply might faulty. according the HP mannual's suggestion, I swapped the power supply tubes with the know-good HP400, no improved again. and swapped into the good HP 400 perform just as well as the originals. 

7. back to basic, I checked each test point's voltage spec, there are 10 to 20V offset in several tube's grid, bad news is the known-good HP400's tube had the offset too.

8. base on the sympton i've notied, I suspect the ripple was the culprit, since there is ripple and capacitor seems fine, the possible reason might be in one stage of the amplifer, its get noisy somehow. So break the feedback loop, testing each stage's output.

9. I found in the stage before the final one was suspicious, its output obviously oscillating, there are stonge output!

10. carefully checking the stage, replace bypass capacitor, changing tube, and do everying i can do for it.

11. I can not stop oscillating of that stages, for few weeks fight, i finnaly reallize, the over all gain is huge, i got break inter stage's link to make sure it was caused by itself instead of the cascading amplifer's hugh gain. 

12. This confirms that if you break the inter-stage connections and shunt each stage’s input with a high-value resistor (e.g., 500k to 1MΩ), all stages become quiet and stable. But once again, it leads to a dead end.

13. I needs close the feedback loop, and do everything again, no progress for few weeks.

14. fine check the ripple voltages on each tube, on power supply, all seems as fine as known-good hp400

15. trying recover the heater's voltage and for each tube, find a faulty tube, test the tube on tube tester, it reading FAIL, but it works well on the known good HP400. i have to replace it with good one anyway.  but no improvement at all

16. while I tried make the hearter supply fit the spec, i found the riplle voltage on the heater supply is super high, 12V supply have a ripple around 1.5Vpp and 6.3V test point have ripple volate rache to 80mVpp. 

17. parrelled new 4700uF with the old one for the heater, the residual reading decreasing never did before.

18. After carefully checking the device’s reaction, I found that the shield of the old capacitor was in poor condition. Tapping it and tightening it with pliers reduced the residual reading by half. 

19. i saddly find that use the old original capacitors is not that bad afer i shocked it. adding more new capacitors does not do any better. 

20. A much worse finding is that the known‑good HP 400 has the same high ripple on the heater, so the reipple is not as bad as I had previously assumed

21. i desoldering and test each filter capacitor on bad one, and compared it with the corresponding one on the known-good HP400. the capacitor on bad one identicaly havd a doubled ESR value.!!

22. Testing a good modern 20 µF–47 µF capacitor shows a D value as low as 0.01 and an ESR of less than 1 Ω. All the capacitors in the faulty unit are uniformly degraded. Every single capacitor has degenerated somehow, and what’s remarkable is that the extent of the damage is quite uniform, it foo me at fist place when i tried to find a obviously bad cap. 

23. By following the heater supply service notes, I identified the bad capacitor in the heater supply — that’s the main reason for the residual reading.