H3000 help? Getting closer, but still can't pinpoint

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
It should also be noted that there's a big difference between using tantalum caps for power supply decoupling, as in this case, where they can and will fail short, and using them for signal coupling, like in the Neve 1073 et al. Has anyone encountered shorted tants in the latter application? I don't think I have.
As far as I remember I think I’ve only encountered shorted Tantalum caps in PSUs rails, but I replaced so many that I might not remember.
I replace them anyway when servicing Neve BA283 cards, also replace the Electrolytics just to be safe
 
Last edited:
Any disadvantage to using aluminum electrolytic caps to replace the tantalums?

plenty of 10uF with 2.5mm lead spacing seem to exist

I don’t see any disadvantage, it might have been different 30 or 40 years ago but modern Electrolytic capacitors are much better than what you had back then.
I actually prefer to have Lytics in power rails, at least they don’t normally fail short circuit
 
Way back when, low ESR electrolytic caps weren't as easy or cheap to come by.

If you've already got it open, I could get behind replacing the tants with low ESR, long-life electrolytics from Panasonic or Nichicon and rest easy knowing I won't have to touch them again for pretty much the rest of my life.
 
I've got an H3000 that a friend found in the trash. It powers on and the display is normal--it loads patches, soft keys work. Passes audio in bypass. Input signal does not meter, and doesn't pass audio when engaged.

After much digging, something is pulling the +15v supply rail down, dragging the +5v supply rail with it (both show only a few fractions of a volt at their output). The -15v and -5v rails seem perfect.

View attachment 89330

Nominally +18v does appear on pin 1 of U7, so the fault appears to be downstream of the regulator.

I've replaced U17 and tantalum C71. The fault still exists, with pins 2 and 3 of U17 showing a short, but would welcome other leads/insights.

For testing, I removed C72 and U20, and the fault persists--this leads me to believe that the +5v portion of the supply is likely fine (is that a fair assumption?). I reinstalled C72 (ceramic) and a fresh 78M05 (both old regulators tested normal once removed from circuit).

My next step was to identify on the schematic all destinations of the +15v supply to look for faults. My preliminary list included 22 ICs and about 15 tantalum caps (I focused on these components as most likely to cause a short).

Some of the ICs are socketed, which made testing without those easy. No result, though--the short persists. I left the unsocketed ones alone for now.

Proceeding to the Tantalum caps, i went through and measured across them in-circuit looking for shorts. Several times (three in particular: C129, C118, C106... schematic excerpts below) showed a short in-circuit, but once removed were verified good (and the short persisted across the empty pads).

As part of basic service, I replaced the large electrolytic filter capacitors, and since I was having issues I replaced the Schottky bridge (CR1) and the 1A bridge rectifier (I knew this was the longest of shots, but since I'm having power issues and I had the parts on hand and it was easy).

On my first examination of the unit, I noticed that a tantalum capacitor in the -15v supply (C74) had blown itself apart. I replaced it before even powering up, and the -15v rail seems absolutely fine, so this is likely unrelated... still, I thought I'd mention it.

I also noticed that at least one of the Murata filters is leaking its dark brown goo all over everything. I cleaned up the board as best I could... should carefully removing these rare/unobtanium parts be my next step? Is there a way to test non-socketed ICs for shorts without removing from circuit? Or is there something else I should try next? Thanks in advance!


View attachment 89331View attachment 89332View attachment 89333
Here's a sort of a club-and-spear approach: connect a DVM, in continuity mode, across the power rail that is shorted. You should get a fairly low reading, like, a few ohms or less. Then, begin removing components, beginning with the most likely - like, any tantalum capacitor, for instance. When you strike gold, the reading on your DVM will jump to a much higher value.
 
Here's a sort of a club-and-spear approach: connect a DVM, in continuity mode, across the power rail that is shorted. You should get a fairly low reading, like, a few ohms or less. Then, begin removing components, beginning with the most likely - like, any tantalum capacitor, for instance. When you strike gold, the reading on your DVM will jump to a much higher value.
Too late. ;-)
 
As a designer of automotive ECUs, about 20 years ago I quit using tantalum capacitors, though for the automotive temperature range I used to derate them to 30% voltage rating, so for a 5V supply I would use a minimum of 25 volt capacitors. However now I use SMD ceramics in 10uF or even 22uF if I need to, and for electrolytics, I use the fairly new polymer hybrid capacitors. They hold their low ESR to at least 40 below zero. I probably would either use high value ceramics or very good (low ESL/ESR electrolytics or polymer hybrids), but also I’d make sure that both ESL and ESR are not a problem for either voltage regulator stability and also on digital supply rails. Especially with CMOS circuitry where very fast bursts of current, often only a few nanoseconds long, tend to flow on clock edges. this would apply to the 5 volt bus on the H3000. Usually there should be a lot of ceramic caps on that rail to help.
 
This thing fired right up today. I replaced all the Tantalum caps with 35v tantalums, except the 1µF tants (which got 63v Wima MKS film caps that had the same 2mm lead spacing) and the 47µF tant (which got an aluminum electrolytic @100v).

I also performed the modification from the technical manual to improve sync of the three processors--two different resistors, and changing the counter ICs to SN74ALS163. One had strangely already been done (and socketed). I installed another socket and did the other, and the system initially wouldn't boot.

But it turns out the first IC I tried was bad somehow, and when I replaced my replacement, it fired right up and has been working great.

Happy with this--just ordered a set of EPROMs to upgrade this to all possible patches through H3500, and then I'll call it a victory. Thanks for the help and advice!
 
It should also be noted that there's a big difference between using tantalum caps for power supply decoupling, as in this case, where they can and will fail short, and using them for signal coupling, like in the Neve 1073 et al. Has anyone encountered shorted tants in the latter application? I don't think I have.
I've had the same thought recently, i've been servicing lots of old neve stuff and I don't believe I've ever encountered a bad signal path tant. Loads of them in our neve consoles, replaced lots of electrolytics but no issues with tants. great application for them it seems.
on power rails however... have encountered a few bad ones on other gear.
 
Back
Top