jokerman777

😵 Spend the last few days trying to adjust the battery and no luck I first tried the method of setting voltage across R42 to 6.55V, keeping RV1 centered and adjust RV2. As I slowly bring things into conduction, voltage across R42 starts slowly decreasing till it get to around 6.8V then it starts jump around 6.55V, as I adjust both RV2 back and forth, the voltage across the resistors seems always likes to stay a few hundred mV apart around 6.55V, and the best I can do is something like 6.4V and 6.6V. In this case negative battery voltage stays at -548V, while I can adjust RV1 to bring positive voltage to ~200V, output DC is wild (somewhere around 80 to 130V) and no small adjustment can bring it close to 0, only can be reduced by bringing positive voltage back up from 200V with either pot. -------------------- I then reset all the pots back to center position and check the overall behavior. When first power up batteries are not conducting with many of the LEDs not lighting up. output voltage is simply 250V/-560V. While keep RV1 (2k) centered, turning RV2 (10k) I get the following behavior (time aligned): Negative voltage: -560V -> -543V ->-549V (stays) Positive voltage: 250V -> 245V -> keep decreasing offset: both positive -> closest to 0 (<+-3V) -> keep increasing as negative voltage (~-100V as + voltage approaches 200V). RV1 didn't really do much IMO, I can use it to trim down positive voltage too just way less change per turn and offset behaves the same as when I turn RV2 to do it. So offset wise it really likes -543V/245V the best and trimming it any closer to 740V it just keep getting worse in negative direction. I measured R42 voltage in this condition and it's 6.9V/7V, which makes sense bc below these values it starts jumping around 6.55V as I observed previously. I did this measure on left channel but I briefly tried right channel as well and it seems to have similar behavior, so likely I made the same mistake in both channels. I feel the servo is working as +/- output offset follows each other closely as long as neither is too wild (>100V). Parts I used are 2sa1413/ksa1156 for 2sa1486s on board and on heatsinks respectively, tubes are supposedly matched JJ E88CC and new production Mullard EL34. Is there any insight on what could be the problem or how to track it down? Many thanks!

So I'm trying to adjust batteries on one channel for now, and was trying to follow Kerry's method. I can set the voltage across the two 22k resistors to be around 6.5V by turning the 10k pot fine, but then it seems turning the 2k pot only affect the positive (~200V) battery voltages and the negative voltage sits still at -548V. I tried turning away the 10k pot a little bit too and neither does it seem to affect the negative voltage but it does change the positive as well, so for now I can have +200V/-548V for this channel. All LEDs in this channel currently light up fine, but weirdly sticking probes across the 22k resistor can dim the some LEDs and cause some squeak noise on the tubes.. Is this behavior expected at all that I shall just move on and adjust the rest or does it look like something is off?

That's what I was told by Bud when I contact them to buy something yesterday. Possibly if you want something in large quantity you can reach out and maybe get a discount over the listed price.

thanks! Never thought of this obvious option, not so resourceful I am : )

Found the rs export site here: https://export.rsdelivers.com/?fromAmericas=1 But when I enter US as the destination country they redirect me to the US site with "Due to local tax laws we request that you place your order with our US website". dk why this is so difficult in 2023 😃 btw I was told bdent is going to close down in Oct, so maybe last call for something.

Hard to find FQP8N80C replacement now as it's experiencing some long lead time. I bought off the 3pcs stock of FQPF8N80C from bdent, and also have ~10pcs of STP8NK80ZFP I bought originally as backup plan, both comes with isolated package and has lower overall power consumption rating. On the surface for now I only need one to fix the +500V/bias supply first, I'm inclined to just use STP8NK80ZFP as I have a slight OCD of if replacing one works I want to replace the other 3 too.... I had success using these in the KGSSHV PSU, or is there good reasons that I shall stick with the Fairchild FQP device?

I had previously considered ultrasonic cleaner for the boards but curious would it mess up with the thermal paste applied? Anyway good luck and hope everything will work out fine! 🙂

Thanks for the suggestion yeah I will probe the voltage drop across them next time I probe the resistors. maybe the line voltage plays a role here too? My trafo primaries are spec'ed at 115V and I believe my mains is above 120V on average, also need to check out how accurate that power measurement thingy is that I borrowed from my friend.

Thanks, it seems my PSU consume twice amount of power as yours.. powering up only all negative rails it's already 22W, adding +250V it becomes 33W so I expect when I get +500V/bias working it will be 40~50W The exploded 10m90s were Q2 and Q3 (Q1 was unknown). I think the shorted D3 is the cause of Q3's death, as for Q2 I don't know exactly yet, possibly the FQP8N80C (Q4) is also dead from the first time failure due to shorting of Q1.

I just went through a couple more rounds of PSU trouble shooting. First time replaced 4 10m90s in the +500V/bias supply where short to bracket happened to one of them, I admit I didn't measure other components other than rectifier diodes, power on and 2 of the 10m90s exploded. (before explosion I was measuring bias voltage and it reads fine) Took it apart and measured more things - D3 zener was dead to short, didn't know if it's the cause or the consequence of somethings else. I replaced all 10m90s again plus the dead zener, op27 and ref102, leaving the hard to source C3675 and FQP8N80C in question. Power on with no load again while monitoring power consumption - in-rush reaches 200W and didn't drop within a short period as I expect it to so I quickly shut it off, something smells funny too. In both rounds I powered on all other supplies individually beforehand and output measured fine. I think I'm going to take it apart again and measure more thoroughly including all the resistors and likely will try just replace all the semiconductors in that rail once I find the harder to get replacement parts. I know it's not the right way to test a device but I measured D-S resistances of the fqp8n80 in circuit and it gives me a stable ~6Mohm which is different from the same device measured in other rails, so something is funky around there. more dummy questions: Is passing forward bias test enough to say that zener diodes are fine? As I was monitoring power consumption of the PSU with no load, each HV rail seems to consume about 10W when stabilized, is that more or less expected? my transformer still lightly buzzes.

Wanted to buy some FQPF8N80C seems uk rs-online only let me enter UK address, is there a way to get them ship worldwide?

Thanks for the all the explanations and the pic! For grounding I wasn't talking about connecting PE and signal ground again in the amp box, ik they should only meet once whether through a breaker or not no matter how many boxes of an amp. I was trying to understand why it's better that they met in the PSU box instead of meeting in the amp box at signal ground, I don't have good reasoning for the later other than I did it that way in single box amp and in my imagination the former extended ground loop between source and amp by the length of umbilical. I will do the former with the advice given, just purely curious, and maybe it doesn't really matter all that much between the two 🙂 In terms of Kevin's pic, I saw a resistor lead coming out of one of the circuit ground terminal block, wasn't clear where it goes to but based on what George said I suppose it goes to rear standoff slot which connected to the chassis, then returned to IEC earth as well as pass to amp box as chassis ground. Idk if it's bc my board is a newer version or not - the pad on that standoff drillout hole is already connected to the on board circuit ground, seems like I just need to put the standoff back there and make sure it's also connected to the chassis I'm all good? rest will be what I already have here: IEC earth, chassis, magnetic shields, output to umbilical. Good point! Craig described the same to me in an earlier post that why I picked this insulation washer based on the length of its neck. I will double check when I re-mount everything 🙂

As for torque I gotten a torque screwdriver and copied homework from one of the early post here and set it to 0.82Nm Sounds good, I will re-inspect all the mountings on the brackets I think I also do have a slight misalignment between the brackets and the semiconductor pins

Thermal compound I used this: https://www.amazon.com/dp/B09VDLH5M6?psc=1&ref=ppx_yo2ov_dt_b_product_details I believe pretty generic kind for CPU mounting in PC. For those 4/40 screws yeah fitting is probably marginally fine, most of them slide in and out of the 7721 freely, some minority samples I feel resistance try to slide them in and I don't use those, I suppose excessive thermal compound filling the gap doesn't help neither.. What did you suggest that could potentially be the cause of trafo buzz? I'd think it's not bc of shorting since over the wknd everything was fine with no shorting and I still heard a buzz. (my trafos are at smaller size and I was given a VA rating of 84, I think SumR/toroidy uses 100VA rated bigger core). Thx!

I took the PSU apart today for simple checkups. On the positive supply side it was the case of 7812 and Q3 10m90s both shorted to the bracket, burning up the insulation. I think I initially had applied force to drive the screws incorrectly such that they bite into the insulation washer and damaged it. They been measuring fine for shorts until I mount the board into chassis and heatsinks and also move it around which probably eventually break the limit. Diodes all measured fine after eliminating shorts but I also ordered 512s for backup. Think I'm going to replace all the 10m90s at the 500V/bias rail later on and see how it goes. Still don't know what was with the spark on the other side of the board.

Mind explain why it should be this way? My naive thought was in the case of single box amp, source and amp have their circuit/signal ground connected at signal input jack of the amp and then both meets mains earth in their own box, thus having amp circuit ground meeting earth at input jack minimizes the ground loop between amp and source, for two box T2 if earth meets circuit ground within the PSU box this ground loop would be extended by the umbilical cable?

Thanks for the suggestions! I will take it apart and do connectivity & diode test first, then maybe reconnect rails one by one for measure. 7812 the case is ground pin so if that's shorted to the bracket it will short signal ground to chassis ground and circuit will still work, this is the only reason I can think of rn that can short the two GNDs. Negative rails was working fine until I see the spark which happened on the negative supply side so I'm not sure now, something might have gone up to smoke. great tip that if I have one leg of the rails working I can use it as reference 🙂

Oh I got parts referencing a BOM where it says stth1512D that is obsolete and I used STTH15S12D instead, I suppose not going to make a difference?

Thank you both for the input! Yes I thought of if chassis/earth ground need to meet signal ground best place would probably be at the signal input, so for now in the PSU Chassis I leave signal GND floating and mains earth/chassis gnd/transformer shields meets at one point and then will be passed to amp chassis ground via umbilical. ------------------------------------- So yesterday I put the PSU together, did connectivity test for shorts and then power on with no load, all output voltages measures fine. Today I thought I want to mount the power transformer tighter to see if it helps with HV transformer buzz so I replaced bolts with longer ones and added lock washer. Power on again with no load (and I didn't do connectivity test as I assume things were same as yesterday), everything was fine until I done measuring the negative voltages and about to go measure the positive voltages, something suddenly sparkled (in the area of the red circle in the picture, too fast I didn't catch which component) and then a bad smell came out, I powered it off right away and nothing looks burnt visually for now. Things I know so far: -Chassis/earth ground is now shorted to on board signal ground, either through the case of 7812 or I might have a short somewhere else. -The anodes of the 10m90s at the +500V/Bias supply side are shorted to gnd, so either one or more of these might have the case shorted to the bracket or I have dead diodes causing it. -On the other side where the sparkle happens the cases of all the transistors are not shorted to the bracket. -Disconnecting the HV trafos power on again +-12V supply still works I know this is way too little information to find out what went wrong other than I have one or more things shorted, I will take it apart further inspect this wknd When testing without load is it ok to run the PSU without heatsinks installed? As that will save quite some effort since I likely need to do multiple iterations of trouble shooting.

Should signal ground meet chassis/earth ground in the PSU box? I found that the center stand-off pad near the output terminals on the PSU board is connected to PSU board circuit ground and mounting a standoff there will make the chassis gnd to signal gnd connection, otherwise the two grounds will be separated in both boxes. I found Kevin's old post in this thread mentioning "everything floats" so should I leave out this stand-off?

I just got the crimping contact version of the connectors earlier this week. Yeah the tools are very expensive, gotten some rusty secondhand one from ebay and hope it will work out... All that is just bc I'm pretty sure I don't want to deal with soldering 8*14 closely spaced soldering cups while also worry about insulation 😅

don't know about anything else but the BIAS is probably fine, measuring after 5M resistor the finite impedance of the voltmeter forms a voltage divider with it.

hmm I've gotten ceramic straight pin tube sockets from thetubestore and turns out the 8pin octal ones don't fit the board. Are these from diy hifi supply the right ones to get? https://diyhifisupply.com/tube-sockets/premium-teflon-sockets/ Searched older posts where Kevin mentioned which ones to get but links are all expired.

Thanks! I got the ones with 0.79mm extension according to a bom posted in an earlier thread here, I guess that's too short to cover TO-220 pads with varies thickness. I will get another batch with longer extension. Actually why is it bad for the extension to protrude the isolation pad? Assuming it's not so long that it will also protrude the metal bracket and board underneath. With that in mind seems like the 7721-10PPSG from the datasheet is a good fit with 2.41mm extension going thru entire TO-220 pads and about half way into the isolation pad.

Thanks for the tips Craig! Metal screws + shoulder washers + thermal compound it is then, no more fooling around. Will for sure check if the length of the shoulder washer from the other side for each installation. 🙂