livewire Posted March 16, 2011 Report Share Posted March 16, 2011 (edited) Some observations and ramblings follow. I've been pouring over the datasheets to better understand the operation of the mosfets in this circuit. I see that the 10m90s that is ahead of the pass regulators is configured as a current source regulator. When switched on by the battery, the 2sc4686a transistor feeds the gate/kathode of the 10m90s as well as the gates of the paralleled fqpf8n80c series pass regulators. It seems that the pass regulator is not "turning on". My questions are: 1) what should the output voltage of the battery be to the base of the 2sc4686a? 2) what should the three mosfets gate voltages be? (as supplied by the 2sc4686a) 3) is there any way to bypass jumper parts of the circuit to get it to operate? 4) am I looking in all the wrong places? One last observation / question for now is the negative rail is only outputting -15 ~16VDC. Is this a intended minimum "leak-by" voltage to drive the battery? I'm really not getting how this battery thing works. The mirrored, cross-connected, cascaded transistors that comprise it are throwing me a curve ball. Thanks in advance for any suggestions, answers or insight that any of you gurus may have.kgsshvpower3(final).pdf Edited March 16, 2011 by livewire Quote Link to comment Share on other sites More sharing options...
livewire Posted March 16, 2011 Report Share Posted March 16, 2011 Is this the same board that got fried twice? I know you don't want to hear this but if it were me I'd repopulate on a new PCB since it was working before it got fried so that means there shouldn't be any errors in the PCB, pre-fry. And since the positive and negative rails are basically the same, other than deriving the bias from the positive rail, if you can't spot the problem on sight it may take a really long time. If you don't want to go that route, maybe post up some pics? Yes it is the same pcb that I fried twice, and the only one I have left. I'm almost to that point in this build. Not ready to go there yet, I still think that if I glean a little more knowledge maybe I can localize the trouble spot. Quote Link to comment Share on other sites More sharing options...
livewire Posted March 17, 2011 Report Share Posted March 17, 2011 Ok, after studying other similar power supply designs, in order to continue with psu troubleshooting, I think that a viable workaround has been found. Im gonna remove (for now) the 2sc4686a sand and the battery sub-circuit in the negative rail and replace it with a zener diode string. (150v-150v-150v) Hello (again) Mouser! Will update in a week or two... Quote Link to comment Share on other sites More sharing options...
tims Posted March 19, 2011 Report Share Posted March 19, 2011 Hi Livewire, I've been following the kgsshv thread from the beginning, so just thought I'd say thanks for your efforts in building this. Will this be the first working prototype, as far as you know? Tim Quote Link to comment Share on other sites More sharing options...
livewire Posted March 20, 2011 Report Share Posted March 20, 2011 (edited) Hi Tim, My efforts have been made possible by the owner of this thread, Doctor Gilmore. Yes, mine is the first prototype as far as I know. "Working prototype"? No, not yet. I have a ways to go. There are at least two others that I know of who are building the KGSSHV at the moment. They have not been posting much here, so I am not aware of their progress. Edited March 20, 2011 by livewire Quote Link to comment Share on other sites More sharing options...
livewire Posted March 24, 2011 Report Share Posted March 24, 2011 I'm back in the hunt, the negative volts are a flowin. The zener V-ref string did the job. So, el problemo is somewhere in the battery circuit. Will order more sand for it from Bdent and ebay. Previously, I had replaced the 2sc3381 sand, but not the 2sa1486. Those checked out ok using a "double diode static test" on my ohmmeter. Due to my ignorance, i wasnt able to check them for voltage under power the last time other than doing a comparo with the positive rail. They didnt match too well. I wasnt able to pinpoint it otherwise, then "ZAP". Well you know that story. Time to button this thing up. (or is it time for "Chapter 2"?) The two amp boards are still an unknown quantity. I did power them up once in the distant past, all the LEDs lit. Still need to test for output, do the zero bias, then connect up the stats. Quote Link to comment Share on other sites More sharing options...
livewire Posted March 24, 2011 Report Share Posted March 24, 2011 Hmmmm. Just found a couple of fried resistors in the battery circuit. Both of the metal film 100 ohm jobs. No visual damage. They are not open either. One now reads 275K ohms, the other 5meg ohms. I thought that I had checked them before when I previously replaced the 2sc3381 sand. Guess not. They have one lead lifted now (desoldered), they would not test correctly for resistance while in the battery circuit due to the output capacitors interfering with the ohms reading, or some such thing. Quote Link to comment Share on other sites More sharing options...
Craig Sawyers Posted March 24, 2011 Report Share Posted March 24, 2011 Hmmmm. Just found a couple of fried resistors in the battery circuit. Both of the metal film 100 ohm jobs. No visual damage. They are not open either. One now reads 275K ohms, the other 5meg ohms. I thought that I had checked them before when I previously replaced the 2sc3381 sand. Guess not. They have one lead lifted now (desoldered), they would not test correctly for resistance while in the battery circuit due to the output capacitors interfering with the ohms reading, or some such thing. Because of the high cost of 2SC3381, why not replace with two separate transistors during fault finding? Two transistors on the same bit of silicon are only necessary to ensure thermal tracking (ie that the Vbe tracks with temperature) Quote Link to comment Share on other sites More sharing options...
livewire Posted March 24, 2011 Report Share Posted March 24, 2011 Because of the high cost of 2SC3381, why not replace with two separate transistors during fault finding? Two transistors on the same bit of silicon are only necessary to ensure thermal tracking (ie that the Vbe tracks with temperature) Good idea, I havent gone that route because that part was specified by KG in the BOM/schematic, I'm not inclined to research substitutes unless it is absolutely necessary. I had already replaced the 2sc3381 the last two times when I accidentally shorted the power supply. I think the one that's in there now is ok. As far as cost or scarcity is concerned, no biggie. Bdent has them in stock for ~$8 each. They are also on ebay for around ten or fifteen dollars each. The last time i bought a few extras. Quote Link to comment Share on other sites More sharing options...
Lil Knight Posted March 24, 2011 Report Share Posted March 24, 2011 There's a person on diyA that sells the 2sc3381 at much cheaper price, 2$/ea, if I don't mistaken. Quote Link to comment Share on other sites More sharing options...
Pars Posted March 24, 2011 Report Share Posted March 24, 2011 ^ Actually 10 for $15, and shipping is reasonable ($3/$4 IIRC to US). These are GR grade however. I bought some, but haven't tested them yet. Quote Link to comment Share on other sites More sharing options...
livewire Posted March 24, 2011 Report Share Posted March 24, 2011 Well shiver me timbers, those are some good prices! Quote Link to comment Share on other sites More sharing options...
ujamerstand Posted March 24, 2011 Report Share Posted March 24, 2011 The same guy also have some 2sk389s. You can hardly find those anywhere any more... Quote Link to comment Share on other sites More sharing options...
livewire Posted March 24, 2011 Report Share Posted March 24, 2011 Yeah, they are almost as rare as hen's teeth. The prices for those on ebay is becoming atrocious. Wont be long before fake ones hit the market, if they haven't already. Quote Link to comment Share on other sites More sharing options...
Pars Posted March 24, 2011 Report Share Posted March 24, 2011 (edited) Yeah, they are almost as rare as hen's teeth. The prices for those on ebay is becoming atrocious. Wont be long before fake ones hit the market, if they haven't already. They had a group buy for those a year or so ago on diyaudio (2SJ109 / 2SK389) The received product could very well have been 2SC3381s... Edited March 24, 2011 by Pars Quote Link to comment Share on other sites More sharing options...
livewire Posted March 25, 2011 Report Share Posted March 25, 2011 (edited) I'm done for the day. Need some coffee. I hooked everything up and flipped the switch. Yay! No fireworks or Zzzzzzzzzzt sounds this time. All eight LEDs on both amp boards light up, although the LED's on both third stage current sources started blinking. This is the stage with the LED between the two 2sc1968 transistors. (see amplifier pdf attached below) The blinking started to speed up and I thought o_O! Something's gonna blow! So I shut it down and unhooked the left channel. I figured I would save one good amp board at this point if something goes south with the right channel under power / test. I fired it up a second time with just the right channel. Same M-O. Shut it down and unhooked the input attenuator. Switched back on for a third time and no more blinky-blinky. So I need to look at how the volume control is wired to the input. With the LEDs now glowing stable, I was able to do some probing. I noted some voltage fluctuation and imbalance between +/- 450V rails. I need to get that under control before doing any biasing of the amp board pots. Under load, the +450V rail is flailing around at 445V to 480V. The -450V rail (with the kludged zener V-ref string) is stable but sitting at 437V unloaded and at 427V under load. Almooooooooost there! I was hoping to hear music through it today. With work in the way, I wont be able to continue putzing around with it until next Monday.kgsshvproduction-final amp.pdf Edited March 25, 2011 by livewire Quote Link to comment Share on other sites More sharing options...
Lil Knight Posted March 25, 2011 Report Share Posted March 25, 2011 Almost there! (that's what she said.) I'm curious how the input wiring could make the LEDs keep blinking. Quote Link to comment Share on other sites More sharing options...
livewire Posted March 25, 2011 Report Share Posted March 25, 2011 Almost there! (that's what she said.) I'm curious how the input wiring could make the LEDs keep blinking. I can only guess. (because I really dont know) Possibly due to oscillation brought on by the voltage imbalance? If the attenuator was hooked up backwards, causing full gain this imbalance might be amplified / multiplied? I'll leave it to one of the amplifier Gods around here to set me straight because i am only speculating and dont have a clue or an o-scope to prove or disprove anything. Quote Link to comment Share on other sites More sharing options...
livewire Posted March 25, 2011 Report Share Posted March 25, 2011 This evening I couldnt resist trying to stabilize the voltage regulation on the +450V rail. Here again, I suspected the active battery circuit was the culprit. I disconnected the amp board load from the psu and tested the voltage. It read +450VDC, steady as a rock. I then down regulated the (+)voltage with the trim pot on the psu to match the voltage output on the negative rail and then hooked it all back to the amp load. The positive rail started to swing wildly again, it was noted the LEDs on the amp board were flickering ever so slightly. So I did the zener string battery replacement on the positive rail as well. Problem gone, although I may up the zener values in the future. Both rails are now at ~427V. So the moral of the story is: I dont like active batteries, active batteries dont like me. YMMV. Quote Link to comment Share on other sites More sharing options...
livewire Posted March 31, 2011 Report Share Posted March 31, 2011 So my struggle continues....more observations below. The power supply is working fine now. I also checked my attenuator connections, they are correct. If I ground the amp inputs (as in zero volume / resistance on the attenuator), the LED in the third stage current source shuts off. (see attached amp pdf below) When I turn up the volume control about a third of the way,(about 15K ohms above ground) the 3rd stage LED lights up. I also tried unhooking the volume control completely and ran wires from the amp pcb inputs directly to the xlr input jack and applied a signal from my DAC.(variable strength signal with the DAC's digital volume control) That also shuts down the 3rd stage LED. The third stage LED stays on with nothing hooked to the amp inputs.(except for the onboard 500k resistors to ground.) I performed the +/- output zero bias procedure, the amp board did respond to adjusting the RV2 trim-pot, but would then drift as much as 10 volts one way or the other. The +output to ground bias was a similar failure. With the RV1 trim-pot pegged to the stop clockwise, I could only get it to within 10 volts above ground (0V). The reading would then drift upwards to about 20 volts. Letting the amp warm up more and re-adjusting had the same effects. At this juncture I suspect that I have some bogus input FETs (2SK389) I noticed that the front side of the plastic package has been abraided, maybe relabeled fakes? I bought them on ebay back in November 2010 from a chinese vendor. I paid $12 each. Today I picked up some 2SK170 sand. Gonna try them and see if things get better. Stay tuned!kgsshvproduction-final amp.pdf Quote Link to comment Share on other sites More sharing options...
El_Doug Posted March 31, 2011 Report Share Posted March 31, 2011 good luck with the new transistors - seems like you're super close to finishing! Quote Link to comment Share on other sites More sharing options...
ujamerstand Posted March 31, 2011 Report Share Posted March 31, 2011 Good Job! Just a little more and you're done! Quote Link to comment Share on other sites More sharing options...
kevin gilmore Posted March 31, 2011 Author Report Share Posted March 31, 2011 I have the power supply done, no problems, and one channel done, also no problems, i'm using a real lsk389. Doubt that matters. An oscilliscope is your friend. especially if the trouble you are having involves an oscillation of some kind. If there are no parts flying across the room, you have to be pretty close. Quote Link to comment Share on other sites More sharing options...
livewire Posted April 1, 2011 Report Share Posted April 1, 2011 Thanks guys for the good words. Thank you Doctor G for bestowing your latest SS design on me. FINALLY! IT LIVES! OMFG I CANNOT BELIEVE IT! After three months of blood, sweat and tears I can hear music through my Stax on my KGSSHV. The 2SK170 sand did the trick. The +/- bias sits right on zero volts now. I still need to do a little tweaking on the +out to ground zero bias. Pics to follow in a few days... Quote Link to comment Share on other sites More sharing options...
Lil Knight Posted April 1, 2011 Report Share Posted April 1, 2011 So my struggle continues....more observations below. Good to hear the PSU is working fine. Did you get the 2SK389s from the the guy 'tubeseller' from ebay? I got some caps from him, all are fine, not sure about other stuffs though. I already finished populating all the boards and also done with the chassis so hopefully I could catch up with you soon. Got a stunning looking attenuator today too Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.