jamesmking

It lives! the son of the mostly modern (RIP) T2 was powered up for the first time this morning (not in a case and with cheap valves) and looks good. All virtual batteries within 0.6V of 740 volts with no adjustments to the trimmers. DC offset after warm up <4V without any adjustments to the trimmers. DC ballance <0.03V power consumption as expected. I did have a bit of a scare. When powering on with my variac for the first time the led next to the EL34 was very dim compared to the others and all the leds where less bright than my original build. Then I remembered that I used different leds in this build and they are in general less bright. I don't think I can be bothered to remove these leds and replace them with the same as the RIP T2. Latest update in the rip T2 troubleshooting, diode testing and comparing the silicon to the son of T2 it looks like one of the J79s is dead... which is a pain since I have no replacements. This dead J79 is on the same channel as the failed 6922. The op-amps and voltage references tested fine. This does not explain why the other channel also had a massive ballance issue between + and - outputs but at least its some progress.

i clean before applying paste. I also don't want to contaminate the cleaning solution. The aim is to get rid of the flux residue, which can be acidic and potentially can breakdown and conduct at the sort of voltages we have.

I moved flat 3 weeks ago (downsized - long story) and the T2 is literally the only stax amp I have on hand here and is my only source of sound! (The mini T2s and blue hawaiis are in storage. Megatron still has no case for the psu). So its a priority project. This is exactly why I have a policy of either building two or each amp or keeping enough spares in stock to build a second one. (The reason I originally did not build a second mostly modern T2 was I wanted to build a fully modern T2 using alternatives for the 2SJ79 and 2SK216 but I had not got around to checking the pcb layout and getting pcbs spun). PCBs are currently drying...

parts stuffing completed on two new mostly modern T2 amp boards. Now waiting for the ultrasonic cleaner to warm up. Then drying, mounting on L bracket and heatsinks... and then *gulp* the stressful part. power on and testing... *fingers crossed* I have no more spare NOS parts, only one 10V voltage reference left, no spare valve sockets etc... so this had better work.

Viva Egoista STX, are you brave enough to listen to it? good news it probably will not sound like crap for long. 💣 💥 bad news you probably will need a new house 🚒 and unless you took the regeneration perk, a new body 🚑

well all the el34s survived, right hand channel input 6922 is bad. Time to go to bed and cry.

RIP MY T2 ever had a day when everything goes wrong and you wished you had not got out of bed.... well I just wrecked my T2. Don't have the energy or will power to even troubleshoot it... what happened. I was slowly over time getting a DC offset on both channels, I would correct with the trim pot until I ran out of travel. My first thought was the valves were getting old. Yesterday I decided to open up on the amp case and check the virtual batteries and voltages.... Well the -560V rail was -600V... I spent most of yesterday looking for shorts, diode checking the transistors and doing the usual troube shooting. Since the psu was basically working but not regulating the -560V I also looked a live voltages... To cut a long story short I discovered partly by logic and partly by accident that even maxing out the trim pot only gave me -9.6V on the 20K resistor on the voltage set string. The voltage reference was correctly outputting 10V. I checked the trimmer and it would change in resistance over its travel. So my conclusion was with the two 540K resistors had drifted up in value and therefore reducing the voltage across the 20K resistor or else something was stealing voltage from the 20K resistor... today I did more testing and desoldered one end of each of the 540K resistors and measured them... 580K and 700K... I remembered that I had replaced all the voltage set resistors with TE 0.1% 1/4W a while ago... I checked the spec sheet and they are rated at 250V.. too little for the job and I think this is what caused them to drift high. My guess is the 700K resistor started to drift high, got more voltage drop across it which naturally increased the drift. I replaced them with Xicon 1/2W 1% 350V and for the -560V rail immediately started regulating again. job done back to listening right.... RIGHT....???? After some testing I connected the psu back to the T2 and the led on both channels close to the -260V terminal did no light. But with no load the -260V rail was rock solid. Connected to a T2 the rail only provided +75V on top of the -560V rail rather than the +300V it was designed to give. I guessed the 2N3904 in the -260V rail had spontaneously decided to fail in the current limiter and so decided to replace it. The 2N3904 I desoldered measured on a peak dca75 as shorted so I was sure I had found the fixed the problem.... again testing without a load everything seemed fine. So back to listening now... right... well... Then I did something really stupid. I thought the psu was off and connected the high voltage cable from the psu to the amp (the low voltage cable with the heater supplies was not connected) there was immediately sparking noises and I quickly turned off the psu. I tested the psu without load - all rails seemed good. I connected the psu to the T2 and powered it up. All the voltage rails where at the correct levels on the T2 But I had massive ballance issues between the + and - halves effecting both channels. All 4 virtual batteries now only read about 550V, (all the leds light so that's something)... SH*T SH*T SH*T DAMN DAMN I guess next step is to test all the valves, take the amp apart and hope the 2SK216 and 2sj79 are ok. I have one complete set of spare 2SK216 and 2sj79 and do have a spare set of amp pcbs so I am debating if it might be simpler just to build a new set of amp boards rather than try to troubleshoot and salvage what I have. today I simply should not have got out of bed.

try https://www.rs-online.com/

they are gone already. radio spares has some https://uk.rs-online.com/web/p/bipolar-transistors/7905413?cm_mmc=UK-PPC-DS3A-_-google-_-3_UK_EN_Semiconductors_Bipolar+Transistors_Exact-_-onsemi+-+7905413+-+MJW21193G-_-mjw21193g&matchtype=e&kwd-303677546319&cq_src=google_ads&cq_cmp=12689234587&cq_term=mjw21193g&cq_plac=&cq_net=g&cq_plt=gp&gclid=Cj0KCQjwiIOmBhDjARIsAP6YhSX25XLs69cZ1m3KLp7BTXoIMRhOSHZejhKgPR-srS1sEjQBYTkcZ3kaAr4TEALw_wcB&gclsrc=aw.ds

personally I use Arctic Silver Ceramique 2 it is non conductive is not metal based and I have had no issues with it so far.

yep, there is a lot of voltage differential across those tabs. I only use STTH512FP so far zero failures zero issues with them.

if the +-12V rails work (with no load some voltage regulators will not regulate properly and most have current and thermal limiters built in) then its highly unlikely the 7812 or 1912 voltage regulators are shorted to ground. with no load I don't see why a working psu cannot be run without heatsinks for short periods of time... however if you have a short on the output side of a HV rail then the pass mosfet/transistor could be trying to output a lot of current and be running hot even with a heatsink... if you have a thermal probe you could CAREFULLY measure the temperatures of the heatsink transistors and see if any are getting excessively hot. when the psu has fully discharged you could run your multimeter in diode check mode on each of the diodes in the psu (both directions). I assume your negative HV rail is working ok. If your psu is symmetrical then the negative HV rail is basically exactly the same as the positive rail in basic topology then you can start making comparative tests between the two. (for my T2 build I did not use the same pcbs or psu as you have so I don't have any direct experience of your psu)

My Stax builds have always had the signal ground completely isolated from the chassis/protective ground - which is connected to the earth pin of the mains plug. amp chassis is earth grounded via a lead going to a star grounding point in the psu chassis - which is where the psu chassis and mains earth meet.

thats close to what I used and should work fine. I used 190Vac 0.18A for the 220VDC rail

gerbers: https://drive.google.com/drive/folders/1r3g2TAtBUaBdiMorTWX7yYgeJ7maQbYW schematics and other things: https://drive.google.com/drive/folders/0B7egryukiT7_TFlEQlBRejdVdDQ?resourcekey=0-nGWwBYQ_Uivj-ciBLYMeaA

Measure the bias voltage before the 5M resistor.

for smd and transistor probing I highly recommend spring loaded very fine tip probes. I use these https://probemaster.com/spring-loaded-micro-tip-test-lead-kit/ the tips are replicable and have an assortment of end types and the leads are high quality... for more hands free measurements I use Sensepeek 4011 PCBite 2x SP10 Probe set https://telonic.co.uk/product/4012-sensepeek-pcbite-kit-with-2-x-sp10-probes-for-dmm/ they are not high voltage rated because of the exposed metal and relatively thin insulation on the wires but they stay in place. The tips are as sharp and fine as the probe master. They also sell scope probes:

im glad you are making progress. checking for shorts from metal tabs on transistors to L bracket/heatsink should be part of a standard test procedure. I personally like aluminium oxide insulation pads... not as cheap as silicon but much more resistance to puncture.. .https://www.mouser.co.uk/ProductDetail/532-4171G

yep I highly agree.. do the simple things first.... miss-wired psu to amp lead or mis-wired chassis sockets is far more likely than leaking insulation.

I am sorry for your loss. Its difficult to evaluate what could have gone wrong, if the psu worked without being connected to the amp then the possibilities are bad wiring between psu and amp board(s) or an issue with the amp board(s)... generally it is a high risk strategy to power up the entre amp to full voltages immediately... during building I hand measure *every* component to make sure it is what I think it is. I have once received different transistors from what I ordered (from a reputable supplier) in bags marked with the components I thought I was getting... turned out to be a picking error... Putting them through a component tester alerted me to the fact they were not what I was expecting. I use an insulation tester to verify there is no leakage between wires within the cable(s) that go from psu to amp. I use a pat tester to verify no leakage from transformer(s) to the psu case no leakage primary to secondary within the transformer(s). before power up I check there is no continuity between any metal tab on the transistors and the L brackets that connect to the heatsinks. I also check the ground wiring very carefully to make sure the amp pcb(s) amp pcbs share a common ground point with all the psu pcb(s) after that I connect the psu to one amp channel only (if possible) and check for continuity between each voltage rail on the psu and the appropriate voltage input on the amp board. This makes sure the wiring to the psu socket, the cable from psu to amp and amp socket to amp pcb is correct. I then do the same for the other amp channel. If the amp connects a voltage rail to the heaters of a valve, I also check for continuity between that voltage rail and the appropriate pins of the valve socket. I prefer to test the psu by itself and run it for several hours not connected to an amp. I bring the psu up slowly on a variac and monitor output voltages and their drift over time. I also use a thermal probe and (carefully) check for overheating components. When possible I power test up one amp channel at a time (again with the psu on a variac) and monitoring voltages on the amp board as I slowly increase the voltage to the transformers. If I don't see all the voltages increasing at about the same rate I abort because a voltage rail is being too heavily loaded - probably by a short.. If I hear the variac buzzing loudly I have a short and abort. I also monitor DC offset and DC ballance on the amp audio output. If this goes well I leave the channel for an hour or so again periodically monitoring voltages, temperatures and DC offset, ballance etc. I then power down, disconnect that channel and test the other amp channel (if possible) before testing both amp channels at the same time. After a few hours powered up I proceed to sine wave testing using a signal generator and monitoring using an oscilloscope. Connecting headphones does not happen until I am very sure the am is stable and reliable. Its much slower this way but provides more opportunity for catching a fault early and less damage if a fault does occur.

mouser and element14 are both reputable suppliers. Your connections to the centre tap input on the grlv is correct . please check the ac voltage from the centre tap to each of the ac voltage inputs. I.e. multimeter from ct (red and orange wire) to the black wire and multimeter probes from ct to yellow wire. This will check your transformer is ok then check the dc voltage across the plus and minus terminals of each of the large input caps. this will check your bridge head rectifier diodes are ok also please take a photo of the underside of the board. If you have a dry joint its possible to get behaviour where you initially get some voltage and then it drops away.

1. check your transformer, you need a centre tapped secondary and make sure the centre tap is going to the CT terminal. If you wire the transformer to the grlv incorrectly you can can one rail working and the other not. 2. measure the ac voltage going into the board (centre tap to one ACV input and then centre tap to the other) to make sure the AC voltage is about the same on both. 3. check the voltage references are outputting 10V 4. carefully check the underside for solder bridges and dry joints. 5. are all the transistors, opamps and voltage refs from reliable suppliers and not off ali express or ebay? - there are lots and lots of fake parts floating around... here is a link to my build and troubleshooting guide: https://www.head-case.org/forums/topic/12269-goldenreference-low-voltage-power-supply/page/27/

😬 either someone just purchased almost 2K in one go or the stock notification I got was an error, or the shipment to farnell got hijacked 😞 digikey still says expected stock around 26th june...

in USA the company is called newark rather than farnell: https://www.newark.com/onsemi/ksc5026mos/trans-npn-800v-1-5a-150deg-c-20w/dp/50H4689?ost=ksc5026mos is the USA link to the same.

the ksc5026mos is back in stock at farnell https://uk.farnell.com/on-semiconductor/ksc5026mos/transistor-npn-800v-1-5a-to-126/dp/2825105?ost=ksc5026mos almost 2K in stock at the moment