Yet another Hafler DH-500 thread...huge current draw with more than 170mV input

bdbell

bdbell

Active Member
I too am dealing with a bit of a nightmare of a Hafler problem. I am doing a recap on a DH-500, I replaced all electrolytic's and most film caps, all small transistors, new multiturn pots. Filter caps were replaced with same - 20,000uF 100v, although I have added another 6800uF per rail while troubleshooting, bridge rectifier was also replaced. And I also replaced and shortened most of the power wires - using 14 gauge mostly. When I started I noticed although the boards were both PC-19c, they had slightly different components - one had 5550/5401's and the other board had ones marked ON128/QP122, there were 3 small films that were 0.1 instead of 0.01 like schematic, and the gate resistors were all 470 ohm on that side instead of 270/470 like other side...I have made them both identical now. For the relay I did not find @ivandezande thread so I ended up ordering an Omron 48v 10amp relay - LY2-DC48

The bias and offset are working properly and set, 90v on the rails, at 100mV input all is good with 0.001% thd, but the problem starts to occur around 180mV input...rail voltage drops 10-12v, scope is all over the place - 200mV it's going really crazy, and at 250mV input it trips my variac set at 5amp trip. Everything is new except the resistors, diode, output transistors and the larger driver transistors. This amp was working fine before the recap and other than the relay and heavier wiring everything else is per the schematic after the rebuild...I find it hard to believe that the relay is causing this behavior but obviously something isn't right - is the relay not compatible or what else could it be?
 
Register to hide this ad
There’s for sure something else going on, but it sounds like you made quite a lot of changes so it will be hard to track down. If it happens in both channels that helps. Is it oscillating?
 
Yeah I did do a lot, but not all at once - I did the filter caps, then tested, then all the electrolytics - tested, then the gate resistors and wiring, tested and then I did the relay and tested - but I am pretty sure I did all these tests at 100mV only, it was only after changing all these did I start trying higher values and seeing the problem. Then I changed the bridge rectifier and all the small signal transistors to see if that would fix the problem - the rails were sagging on load, and the small signal transistors I had read several places should be changed.

It's not oscillating as far as I can see but it's not one channel - pretty sure it's happening in both. I have been using my qa403 to test with and the level is difficult to ramp up/down (QA40xPlot app) - I will use my signal generator and do a test slowly ramping up the input and take a video in the morning.
 
Shootin' from the hip here...

But when you say the rail voltages drop, it made me wonder if there is something limiting the power supply like the transformer not delivering enough current?

Is this something that comes on slowly as you increase the input voltage or is it more of like falling off a cliff?
 
It's definitely more of a 'falling of a cliff' - as I mentioned it was tripping the reset on my variac at 5 amps, I thought that the trip reset on my PR570 might be limiting the current so I plugged it directly into 120v, and when I hit 180mV it blew the 15amp breaker - first time in ~3 years in this location that I have done that!

I have it set up now with input from a signal generator and output to a load and then going into my scope - first video below shows what I am seeing with no input at all - at 1v, looks pretty nasty closer up...second video is with input ramping from 150mV to 250mV and back down. I noticed with this setup it is more stable than with the QA403, and the problem is not happening until beyond 200mV - above 250mV and I see the current draw very quickly rising and pretty sure 300mV is going to trip the variac.



With 120v input it's drawing 1.3amps and the rail voltage is 84v.

IMG_8763 (Medium).jpgIMG_8761 (Medium).jpg
 
It did seem to be more of a falling of a cliff with the QA403 but I think that has more to do with the output from my signal gen not being as hot as the QA403 so it's just occurring at a higher value. And I wasn't able to see the signal as clearly on the app with the QA403 as with the Sigilent - it was totally going all over the place when things started to go wrong, but pretty clear with the scope.
 
^^ I was also wondering this. If the feedback were disabled or reduced for some reason, it may be producing an awful lot more output than expected for the input signal level, though even if it were at full scream it seems surprising it could kick a 15 amp breaker without blowing any internal fuses.
 
Thanks for the replies - after the things I did below I am seeing clean sine with 10.4v p-p output at 250mV input - at 260mV the problem starts with output ~11v.

@gadget73 - yes it was quite shocking that it blew the breaker - I live in a pretty small space so the circuit was probably pretty loaded already though.

I am hearing a clicking noise on my personal amp which is on the same circuit in conjunction with the noise in the signal - which leads me to think this has something to do with the grounding

I did a lot more checking on things I had changed, looked over all the before photos and found a few things - the compensation caps from Drain to ground that were originally installed were 0.01, I had put 0.1 which is what the parts list shows. I also removed the big copper bar I had installed and put in a copper rod similar to original, and I had run one ground wire to the back outputs and bridged them there instead of two, so I changed that back to original. I have also removed the extra caps I had added as they were obviously not helping. Unfortunately none of these things solved it.

I am just about to do yet another check of the values of capacitors I replaced on the driver boards, and if I don't find the problem there I was going to eliminate the thermal cutoff and do a test without it in circuit to rule it out.
 
I have confirmed that I installed the same capacitors that were on the original driver boards - except C12/C14 were increased to 220uF and the input cap C1 to 2.2uF. The only other thing I noticed is the parts list shows 1uF for C6 but looking at a picture of the originals and it definitely has .1 in both boards in this location. Now I'll test without thermal protection but posting this in case anyone else notices anything.

dh-500-original.JPGIMG_8769 (Large).jpg
 
I wired an iec plug at the front where the switch/fuse is and connected it directly to the primary - one going through a CL80 and the fuse - and used a really heavy duty 10 or 12 gauge power cord, same results so it's not the thermal cutoff circuit or the power cord...the search continues
 
Fyi... when a unit is generally working it's often recommended to replace with as-installed parts rather than schematic values as schematics are not necessarily 100% correct in every regard. Can be typos, production changes, sometimes even service bulletins that are not reflected in a schematic.
 
Last edited:
^^That is a good point and I have done that - but as mentioned in the first post the two boards were different and I made them both the same as the one that I thought was more correct. That board had a 0.001uF for C4, C5 and C23 which is what the schematic shows - the other board had 0.01uF here so I went with what the schematic showed - I guess that's a possibility but why it not have been a problem with the one board before, it did also have 470 ohm gate resistors in place of the 220 though!

Guess I will try 0.01
 
OK, so I changed them all to 0.01, but of course Hafler didn't make it that easy, I can see the oscillation in one channel much more than the other but still there in both, happening at a higher input now - looks like the values were selected for each side based on the output transistors selected so I need to get it back to the way it was originally. The thing is I was an idiot and threw out the original caps...really never usually do that until I am sure I am done with a project but it happened...never again.

Anyway - what I have determined is that board 1 had 0.001 for c4, c5 and c23 and 220/470 gate resistors, and board 2 had 0.01 and all 470 ohm gate resistors, and on the p-channel mosfets that had 470 instead of 220 it had a different compensation cap, I can see ??COR and 128T69 in the picture but don't have any other views unfortunately - any ideas?

IMG_8647v.JPG

IMG_8647.JPG
 
Is it the fact that the output mosfets were not all equal (2sk176/2sj56) - that the differences were required in the components, or that the design was bad...just really shouldn't be this difficult!

I figured I will try to focus on getting one channel working properly so I disconnected the left channel and have tried several values for the compensation cap I am not sure about, with a 0.0022 I see the oscillations at 240mV, with a 0.1 - I see them at 290mV - going higher doesn't help...time to take a step back from this one, think about it and wait for feedback.

Really wishing I would have gotten some upgrade PCB's for this - but am I going to have the same problem with them if it's differences in the mosfets themselves that required the driver board mods?
 
It's got all the original mosfets, the n-channel mosfests are marked K176 which is 2sk176, the p-channel is marked QP113 - but the qp113 is actually the part # for 2sj56. I removed all of them - cleaned, tested and put them back with new mica and thermal paste, was very careful to put them back in exactly the same locations.

Apparently Hafler used a lot of surplus parts and mixed and matched things in a pretty non-standard way so it's common to see two different driver boards in the same amp. The pcb's are both PC-19c though and I could not see any differences in the traces at all.
 
I was starting to think that the original slight differences in the components were what were preventing the oscillations I am seeing at relatively low input levels. Seems crazy to have the boards different to do this though, and really not finding anyone else having an issue like this so just not sure what's wrong here.
 
I took the relay out of the circuit - with no load I get no oscillation, but with the load I see oscillation. Just wanted to confirm that it is not the relay, and now that I have also removed the thermal cutoff circuit I have only the driver board differences to suspect.

To recap;

I did a full recap and a few other mods on this DH-500, before I tested it with 750mV input and was getting clean output. After the work I am getting a HF oscillation at inputs of 290-300mV input (originally 170-200mV). I discovered when I started the rebuild that the two driver boards had slightly different components - mainly C4, C5, and C23 and the resistors underneath - the small signal transistors were also different between the boards, but I have replaced them all with new matched pairs. The 2 100uF electrolytic's were upped to 220uF, but otherwise all values of capacitors are original, diodes and resistors (except gate) have not been touched. Original config was;

Board 1(left): 0.001µF C4, C5, and C23, 220/470 gate resistors
Board 2(right): 0.01µF C4, C5, and C23, 470 ohm all gates

As mentioned in post 15 I am unsure of one of the caps that go from the p-channel mosfets to ground (Board2) - all others were 0.01. Other than that I have put it back to exactly the same values as original and am still seeing a 50-60us oscillation at the peak (post 5) of both channels at the same time. In the right channel that has the unknown cap I tried different values as mention in post 16 - 0.0022 caused the oscillations 240mV, and 0.1 at 300mV - higher values didn't help.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom