jamesmking

GR78xx build notes: Writing direction on the opamp and voltage reference match the writing direction on the pcb I suggest checking the led orientation with a multi-meter in diode test mode, finding out which way round the multi-meter leads need to be to make the diode light and then marking either the positive end of the led with a red pen or the negative with a black pen whilst holding the led with twisters - if you don't hold the led with twisters it has a nasty habit of sticking to the pen tip. The leds face opposite directions. Positive and negative ends are marked in the picture with red and black lines respectively. All transistors are shown with their case markings. Its a good idea to solder the leds in first and then check. D2 has the end closest to the Vin pins connected ground and so its easy to check its orientation is correct, the other led is opposite orientation. The DN2540 is a depletion device so if you measure resistance between pins, two of the pins will appear to be shorted - this is normal and shown on the picture. The tab is also electrically connected to the centre pin... The top left pin of the hn4a51j is also connected to the bottom middle pin on the ocb and will therefore also measure as a short. The pmsta devices have 3 digit case codes the first digit being the place of manufacture, the kst devices have two digit codes. Be aware Both the pmsta42 and kst42 have codes 1D but the pmsta42 has a t or other letter before the 1D for the manufacturing location. voltage ref should have 10V between pins 4 and 6 If you test the board by itself it must have capacitance connected to the output in excess of 20uF in order to regulate and not do strange things like increase the output voltage when the load increases. When setup for 15V output, input voltages bellow about 1.7V should result in very little current draw <1mA and no leds lit . At around 1.7V input D2 show glow dimly and increasing the input voltage should result in D2 glowing brighter, the current draw increasing and the output voltage increasing. When D2 is fully bright the current draw should be about 10mA. At about 17.5V input D1 should begin to light . Increasing the input voltage further will make D1 brighter and the current draw increase towards 10mA current the draw when the led is fully bright. At around 18.3V input the output should stabilise at around 15V and increasing the input voltage further should have no effect on the output voltage or current draw. The second 78xx I built required 0.2V more to regulate fully despite having a 8mV lower output. Load testing Expect approximately a 23mV decrease in output for each 100mA of load current drawn from the gr78xx. At 200mA the components on the board should be warm but easy to touch and leave your finger on. The MJW211 should also be a little warm. NOTE the metal tab of the MJW is live so don't short it to something. Voltage drop under load test, no heat-sinking, MJW211 close up against the pcb, 18.3V input. The drop in output voltage is the same using a 47uF output cap as 220uF. 100mA 23mV drop 200mA 46mV drop 300mA 69mV drop 400mA 93mV drop 500mA 116mV drop (note tab of mjw gets too hold to touch for more than a moment after a few minutes of this load. heat-sinking recommended) The second 78xx I built had voltage drops within 2mV of the first. The voltage drop under load is a bit more than a traditional 7815 to-220 voltage regulator under the same conditions. Removing the load results in the output voltage of the gr78xx going back to within 1mv of the original no load output almost instantly, even from the hot 500mA load. the voltage drop is fairly linear with load. Otuput voltage drift under constant load as the device heats up or cools down is very low. LED direction verification for gr78xx lower led D2 multimeter in diode check mode red probe on top of led black probe on bottom (as indicated in photo bellow) you should get about 0.89V drop led will not light. probes reversed meter should say OL upper led D1 multimeter in diode check mode black probe on top of led red probe on bottom (as indicated in photo bellow) you should get about 1.66V drop led will light. probes reversed meter should say OL 79xx build notes All polar caps, Zener and LEDs are in the opposite orientation to the 78xx, All resistors are in more or less the same positions and orientations as the 78xx, all transistors that are xx42xx become xx92xx and visa versa. NOTE in the BOM Q6 and Q4 appear to be the same on both the 78xx and 79xx however, on the silk screen the physical positions of Q4 and Q6 swapped in all other cases the transistors change in the BOM rather than the silkscreen designations moving. The opamp and LT voltage reference are in the same orientation as the 78xx. The HN4A51J becomes HN4C51J and the MJW211 changes from 94 to 93. All positions that hold a KST device continue to do so and the same is true for the PMSTA. Only the type of device swaps from NPN or PNP i.e. from kst42 to kst92 and visa versa or pmsta42 to pmsa92 and visa versa. The 79xx has similar load behaviour to the 78xx. LED direction verification for gr79xx lower led D2 multimeter in diode check mode red probe on bottom of led black probe on top (as indicated in photo bellow) you should get about 0.89V drop led will not light. probes reversed meter should say OL upper led D1 multimeter in diode check mode black probe on bottom of led red probe on top (as indicated in photo bellow) you should get about 1.66V drop led will light. probes reversed meter should say OL

Thank you for the quick reply You are absolutely right about the output capacitance. I recreated the golden ref through hole output cap setup of a 220uF and 4.7uF in parallel and that did the trick. D1 now slowly comes on when the input goes above 17.5v and behaves much more like the golden reference through hole and the gr78xx is now happily regulating at 15.014V with no load. At 18.26v the output is now 15.015v. 50mA load the output is down from 15.015v to 15.003V, 100mA output is 14.991V, 200mA output is 14.967V all with 18.26v input and no heatsinking. Switching off the 200mA load the output is straight back to 15.015V. Playing around a little more regulation works with just the 4.7uF film cap on one of the two boards but results in the other board increasing the output voltage with increasing load! putting back the 220uF caps makes the second board behave like the first with similar voltage drops with increasing current draw. The silkscreen for the gr78xx says hn4a51j which is a pnp part which is consistent with the through hole golden ref which uses 2x ksp92ta which is also a pnp transistor... now if i could get more voltage references I could build the gr79xx boards....

has anyone built a gr78xx and tested it yet? my gr78xx refuses to regulate. 😞 I just finished building 2 gr78xx. tested both for shorts and not connected components. each in turn to a DC power supply, with current limiting set to 50mA, nothing connected to the gr78xx output except a multimeter. Things started well, D2 just barely started to glow at 1.8V input to the gr78xx, as the input increased D2 glow increased... D1 suddenly started to glow at 16.3V input and the output voltage at this point was 15.088V. Current draw from the dc power supply was around 20mA, no bangs no surprises... but when I increased the input voltage more the ouput voltage just followed it. At 18.26 input the output was 17.073V.... I checked the voltage ref 10.005V between pins 4 and 6. No components feel overly hot. Both gr78xx have been ultrasonically cleaned. Both the gr78xx behave exactly the same way. On a DC load the output voltage drops with increasing current draw at 250mA the output is 14.58V BUT D1 and D2 remain lit with no change in brightness compared to no load on the output. any ideas what could be the problem?

I believe there is a small error in the BOM for the GR79xx. Q2 is stated as "PNP silicon kst42 or MMBTA92" but the kst42 is NPN I believe it should be a kst92, given in the golden reference through hole, the transistor at that position is a PNP AND the MMBTA92 is a PNP.

Thank you for the group buy, my boards arrived a few minutes ago.

the capon coated volume volume extender shaft is the fuse.... all my lt1021s from mouser have a 2 digit code at the top right of the chip (most have e3) running vertically downwards... so im sure those voltage refs are legit too. P.S. talking of fakes dont buy from little diode on ebay in the uk... I got 2 sets of 4 2sa1486 from him fake.. C to E breakdown voltage a massive 360V... specsheet 600V... and 10 2SJ79, on a peak dca75 transistor identifier it shows D and S are swapped compared to the spec sheet! I would not want to see what happens with that in a circuit... Breakdown voltage using the specsheet pinout 0V! breakdown voltage using the pinout the dca says <100V... spec sheet 200V... and the "best" one so far from nikkolec on ebay $200 of 2sa1968s. Peak dca75 detects them as npn but spec sheets says pnp! breakdown voltage is 190V not 900V guaranteed firework display if used.... sigh

I'm running + and - 15v, 6922 and el34 from the same transformer all heaters are AC. As long as the insulation on the windings can handle the 400V between el34 and the other windings its fine.

I believe all the 6922s might be run off one winding since the heaters are ground referenced and the current draw is much less than the el34s but I have not tried it. As joamat says the el34s cant share a heater winding with the 6922s since the el34 heaters are -400V with respect to ground. I have not tried running both channels el34s from a single winding because a. thats a lot of current draw 6A ignoring switch on inrush and magnetic losses b like joamat I don't know if it would be safe to do so. c separate windings provides a bit more isolation if/when things go wrong. You might want to message Kevin Gilmore for his advice.... P.S. jose have you signed up to the please add screw terminals to the mini t2 amp boards request yet 👿

paid. good luck everyone with your builds

I have run the amp without the balance servo, just by carefully adjusting the trimmer, but the opto coupler is connected to the dc offset (i.e. make the average of the + and - audio signals 0v with respect to ground) not the balance servo (which makes the average of the + and - audio signals 0v with respect to each other only. also check the valves/tubes if you have a tester... I have found the 6922s to be a bit fragile and the el34s have -400V on their heaters so if the negative 560V rail has gone crazy there could be heater to cathode leakage... do you have a variac? I always bring my new builds up to full voltage slowly... My amp has two transformers one for the heaters and +-15v and one for the high voltages. I connected the low voltage transformer directly to the mains and only connected the hv transformer to the variac. What resistors did you use? I used vishay crcw1206 which are rated at 200v and crcw0805 for the 1/8w which are rated at 150v., they were the highest rated voltages I could find on mouser for 100ppm 1%... I have a dca75 from peak electronics, https://www.peakelec.co.uk/acatalog/dca75-dca-pro-semiconductor-analyser.html#SID=24 its been very useful for post mortem checking (if I de-solder a transistor I can compare the dca75 measurement with a spare transistor of the same type. I often find failed transistors simply show up as dual back to back diodes or dead shorts. Note the DCA does have trouble identifying 10m90s and similar). I have even found some fake transistors using it. Its not designed for in circuit testing, but it can be use in circuit for comparisons between transistors in the left and right amp boards. If you have not got a dca75 I highly recommend it. You can also get an adaptor for sot23 https://www.peakelec.co.uk/acatalog/pca23-peak-component-adapter-sot23.html and I found an adapter on ebay for sot23-6, https://www.ebay.co.uk/itm/SOT23-SOT23-6-SOT23-6L-IC-Testsockel-Programmieradapter-Einbrennsockel-F3I4/383314954496?hash=item593f5a0900:g:dWoAAOSwTYZd8frU so I can test dual smd transistors too. I also created a little op amp test circuit on a bread board, unity gain just so that I could power it and feed in a test signal and scope the output, I got a adaptor for 8 pin smd from ebay https://www.ebay.co.uk/itm/SMD-SO8-SOP8-SOIC8-DIP8-2-54mm-DIP8-Programmier-Adapter-SMD-Sockel-150mil/252912033731?hash=item3ae2bb5fc3:g:yKcAAOSwmE9b-JqP so I could also test smd opamps.... I love having lots of test jigs. Good luck with your troubleshooting and repairs.

For me its not about the bit rates or sample depth but the care and attention put into the recording/remastering process. Very few companies make dedicated recordings in each format, an exception is the label 2xHD which document the recording chain and use separate brand adc for the different formats, DSD vs PCM, and so you can expect to hear differences due to the different ADC as much as the different digital formats. I suspect most multi format releases are recorded and mastered in a single format at a single sample rate and then converted to the other formats. The conversion is not perfect and this is easy to demonstrate. If you take a very low frequency square wave and convert from 44K to 88K or go the other way the square wave is not preserved... conceptually you just remove half the samples or double each sample but in reality its a fast Fourier transform and digital filters (c.f. Izotope RX)... and FFT does not work well at low frequencies and neither do the digital filters.. I hear less differences between modern dacs than I do other components like speakers, cartridges etc. I will say the early dacs were terrible. I have spoken with one hifi designer who admitted that of all the audio technology dacs have probably improved the most in the last over the period late 80s-2010s. certainly I could not find a dac I could live with until the late 2000s and then it was super expensive, now I am happy with a V90 with a golden reference LV psu... A very cost effective combo which performs as least as well as my crazy expensive (even second hand) DCS elgar plus, even if the v90 does not have DSD support, input options etc. MQA is loss-full compression (it was original advertised as lossless but analysis has shown it is lossy) and has not really gained much traction, its potentially full of digital rights management and designed for surround sound. (https://audiophilereview.com/cd-dac-digital/mqa-the-facts-versus-the-fiction.html)... now its advertised as less lossy than mp3 and the DRM is not currently switched on... MQA playback requires licenses and so its unlikely to appear in open source playback software and even Jriver media centre has no plans to add it... I guess the license costs are too high. Software like Jriver media centre can convert DSD to standard PCM and this is how I listen to DSD. The advantage of DSD is that the ADC converter is simpler and has less filtering. The disadvantage is that the playback requires more filtering (DSD generates a lot of high frequency noise) and its not easy to edit DSD... almost all editing is done by converting DXD 384K PCM and then converting back to DSD this process is supposed to incur no losses... There is very little music available in DSD format and even less thats only available in DSD... My philosophy is get a modern dac, make the psu good and use the saved money on a better amplifier/speaker/headphone where there are still large performance differences... and don't forget to clean the connectors with deoxit or similar periodically. My mii T2 was recently sounding off colour, a bit lifeless and lacking detail in the mids and just sort of tried. I cleaned up the rcas and its back to singing again...The connectors on my van den hul first seem to love to oxidise a lot and rather quickly too... the cotton bud came out almost black from a single rca connector... regards and happy listening James

The subcontracting is everywhere. For example the courier dpd in the uk, deliveries are by DPD and if you have a company account with them you can send things. If you don't have a company account you can use DPD online or DPD local... all three are separate companies plus DPD online has no telephone support, not even an support email address. DPD local has telephone support as does DPD. None of the support of any of the three companies has access to the records or support of any of the others and you effectively have separate accounts with each one... what a mess. The DPD website does not mention this but as soon as things go wrong you find out the hard way... especially if you sent something using DPD online and expect to talk to someone on the phone rather than by web text chat to someone in India who has no clue.... Many Amazon deliveries in the uk done by "amazon logistics" AKA yodel. Judging by the internet and my personal experience Yodel is probably the worst delivery company in the uk... I ordered something before corona virus lock down, delivery to my university address. By the time it arrived the university was closed due to lockdown. Yodel did not contact me to ask for a new delivery address or inform me of any problem. Fortunately I was pro active, when I looked at the tracking it said to go to a web page to change the delivery address... when I tried the page said you don't have permission to change the delivery address!. So I tried to contact customer support... Telephone support... closed because of corona virus. online support.. busy waiting time 5 days.... when I finally got through the support person 5 days later they where not sure if they could change a delivery address... 10 minutes waiting their line manager approved the change. I said my home address and they said no that address is wrong and for that postcode Yodels system says "N 40 london" I said thats an impossible address its not even in the correct format. I was informed it was the only address change they could make or send it back to the sender. I asked if they could hold it and I get the sender to change the address? NO the sender can only change the address once its sent back to them.... long story short 2 months later I am still waiting for delivery...

I built two mini t2s, I always build in pairs or they get lonely... there is a bit or board flex when inserting or removing valves, (the ptfe sockets I have grip very tightly). No issues I can see with 1.6mm and 1oz, that's what I have used for all my builds, (4 dual output GRHV boards, 2 single output GRHV boards, 7 GRLV boards, 12 synchronous rectifiers, 4 diamond buffers, 4 blue hawaii amp boards, 4 alpha centauri amp boards and 2 alpha centauri psu boards)..

I have had no trouble using the zip with jlcpcb in china, I have sync rectifiers in my grlv in my blue hawaiis, mini t2s, psus I made for my dacs and a preamp I made,. Please be aware the sync rectifier will not work with centre tapped windings . so if you are making a grLV you need two independent windings and two sync rectifiers,. you cant use the GRLV option for centre tap and one rectifier I have tested it myself the sync rect does not work in this configuration but a normal diode rectifier does. The pin pitch is rather small and you need to check very carefully for shorts between pins 3 and 4, (pins 1-3 are internally connected together) and between the metal base and pins 1-4. I created a small test rig for the sync rect, I used a spare grlv board, put a female header where the rectifier solders in, added a largish value smoothing cap and bring the sync rectifier under test up slowly using a current limited dc supply and when I believe its ok, I swap the polarity of the dc supply and repeat so all 4 psmn040 are tested. I then use a 15v transformer and bring the sync rectifier up slowly on a variac. So far I have had one psmn040 smoke, I think I had a soldering short between the metal base and pins 1 or 2 or 3 or 4. Fortunately the expensive lt4320 survived without issue and I was able to reuse it. the test rig also includes wire loops for attaching a multi meter, scope and electronic dc load. I populated the diode next to the cap and one led partly to help drain the cap and to give a visual indication the board is live. Once the board has passed variac tests, I test switch on with no variac so the cap will cause a reasonable inrush, then I test with an electronic DC load up to 2A. .

joke { 5 digits is too low accuracy, you need at least a 8.5 digit multimeter calibrated monthly in a sealed controlled environment and Faraday cage. The resistors should all be 0.01% 5ppm or better and the leds need to be curve traced from 0V to 5V and matched within 0.01% on current. This should make the LV psu board more expensive and time consuming than the rest of the amp. 🙂 } /joke I don't know if the cfa has any special requirements but for both my blue hawaii and mini t2 builds I went for 1% 50ppm resistors, - koa brand. I did not match the leds other than they where all from the same reel and so had the same nominal specs. I aim for voltages within 1% of the target given I use 1% resistors. The ppm of the resistors will effect voltage drift with temperature, if you want good accuracy and stable voltages build the board with low ppm 1% resistors, measure the voltages, calculate the error and replace just the resistors that set the output voltage. I don't see any point in spending lots of money on better resistors everywhere when I can spend the money I save on music or other vices 😉 ...

I used two transformers. both with electromagnetic shields, static shields and potted from http://www.mueller-rondo.com/kontakt.htm . (The low voltage transformer can be used for the blue hawaii - the 6922 windings are just not used the other voltages and currents are fine.) low voltage: transformer two windings 6.3V 4.5A for EL34 heaters one winding for each channel NOTE these windings are -400VDC to ground two windings 6.3V 1.5A for 6992 heaters one for each channel and my HT delay board two windings 15.5V 0.4A for golden reference LV board + and - 15VDC high voltage transformer one winding 365V 0.2A for dual golden reference HV board -460VDC. NOTE 365VAC gives 516VDC which gives less than 10% margin before reaching the 550V limit of most high voltage 550V 470uF input caps on the golden reference HV... so be careful if your household mains voltages are usually higher than spec. one winding 322V 0.2A for dual golden reference HV board +400VDC and +580V stax bias one winding 192V 0.18A for golden reference HV board +220VDC The blue hawaii just fits in a single 2u case 400mm deep, using 1 transformer with less windings - one less high voltage and two less 6.3V, golden reference LV board and golden reference HV board. The mini T2 has two transformers AND an extra HV board so the issue is not height but depth, to give you a size idea the transformers are Low voltage diameter 105mm height 55mm High voltage diameter 115mm height 65mm not having shields in the transformers will result in some hum and electrical interference if you turn the volume all the way up if you go single case. Not potting the transformers might make them a little smaller. putting the transformers into a separate case makes both the blue hawaii and mini t2 absolutely silent. you cant easily use the space bellow the mini t2 amp boards because they have 4 pillars which screw into the bottom of the case so you don't bend the amp boards when inserting or removing the valves. Neither the golden reference LV or HV boards will fit bellow the am boards because of this. for both my blue hawaii and mini t2 builds I went with the following case https://modushop.biz/site/index.php?route=product/product&path=102&product_id=195 its 2u, 400mm deep and has large heat sinking and is fully aluminium so fairly easy to drill. The heat sinking is overkill for the mini t2 - which runs quite cool, but the blue hawaii needs those large heat sinks, For the umbilical chord I used 1KV silicon rubber multi-strand copper hookup wires (https://www.mouser.co.uk/Search/Refine?Keyword=CT2956) for good flexibility and Russian 19 pin military connectors (https://www.ebay.co.uk/itm/19-pin-Soviet-Military-connector-Female-Male-Set-Oty-1/254196620095?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2057872.m2749.l2649). I then covered the bundle with expandable nylon braid and tested the leads with an insulation tester and they passed 5 minutes at 2500V with 20+Gohm resistance. They failed insulation at 5KV. P.S. if you build golden reference HV boards kemet do 550V 470uF caps which are 65mm high and will fit in a 2u case https://www.mouser.co.uk/Passive-Components/_/N-5g73Z1yzvvqx?Keyword=kemet+550V+470uF&FS=True. Be careful Kemet also sell 680uF 550V caps that are 80mm high and will need a 3u case to fit . (https://www.mouser.co.uk/ProductDetail/KEMET/ALC10A681EL550?qs=%2Fha2pyFadug678O2NPlQ%2F8WMW0dBZWV2y30oREQu6XgezeenvV90cA%3D%3D). for the +220VDC golden ref board you can save some money and go kemet 470uf 450V or even 400V. I didnt simply because I might re-purpose the 220V board to a higher voltage sometime.But the 550v 470uF kemet caps are not cheap. P.S. the golden reference LV board is not necessary, only the servos are run from the + and - 15V lines so you could just populate the low voltage section of the GRHV board. However, if you don't want thumps on switch-off the + and -12V lines require large capacitors so the servos stay powered while the HV lines fade. I found ~15000uF to 22000uF is required but these caps are physically too large to fit in the simple LV section of the GRHV board. They fit fine in the GRLV board, but a full GRLV board is extra expense just to avoid switch off thump. I use GRLV boards in lots of projects so again they can be repurposed...

Thank you. Paid

OneGuy: "both hands free"?.... you also only have two hands? but can the reflow oven do double sided?

joamat: on the quick air flow 5 out of 120 !!! even then you can orbit small diodes if you release the tweezers before the solder has melted the quick is powerful, 120 will almost remove through hole comonents and your third arm.... maybe thats why mwl168 only has 2 hands.... temperature 330C to 360C depending upon ground plane, size of component. solder paste nothing special off ebay its MECHANIC XG-Z40 Liquid Solder Soldering Paste 10cc has a good amount of flux in it seems to work quite well for my soldering technique but then I have not tried any other paste.... mwl168: you DON'T have 3 hands?😕 My smd soldering method requires 2, left hand holding the twessers, right hand holding the hot air nozzle....

mini t2 is not alot more than a bh, a few extra transformer windings - it needs seperate windings for the 6922s and el34s, an extra 220V supply or dropper resistor, extra valve sockets and 4 6922 valves against that smd parts are generally cheaper than the through hole and I found that in many cases - especially the resistors buying 10 of a value in smd was cheaper than buying 6 - the bulk discounts scale much better than through hole. It is true that its possible to build a blue hawaii in a single case and still fit in golden reference LV and HV boards, its not realy possible to do this with the mini t2 so you need a second case and wiring between the cases. However, it is possible to fit all the pcbs in a case the same size as the blue hawaii and just have the transformers in a small cheap second case.... - which is what I have done.

tomislavkufrin There is the mini t2, if you dont mind surface mount for the amp boards, you can use exactly the same golden reference lv supply as the blue hawaii - zero changes, the same golden reference hv supply with two resistors changed to go from -400V to the mini t2 -460V, add another golden ref power supply for the +220V or a dropped resistor from the +400V to save an extra power supply board and transformer winding.. if you use two transformers - one to supply the heaters and the plus and minius 15V DC - that transformer can also be used on the blue hawaii the second transformer can supply the voltages for the 220V, -460V and +400V. I think the sound of the mini t2 is better than the blue hawaii (a big enough difference that my mini T2 is now my daily stax amp and the mini t2 runs cooler) and quite a lot of the mini t2 can be re purposed if you do decide to build a blue hawaii some time. The blue hawaii is also a nice amp if you dont want to do smd soldering. personally Im waiting for someone to design a reliable full diy t2 using modern components.... Kevin? JoaMat?

fuses degrade the sound 🙂

I got the quick based on a review my lewis rossmann on you tube who does apple mac repairs . I tend to be paranoid, so I assume solder paste will go under components and cause shorts... I plan to get some experience just dropping the components on the raw solder paste and applying heat, but I did not want to take the risk with the mini t2. My general philosophy is find something that works and then optimise it. there is one issue with my smd method... if you apply heat with the soldering iron for too long you burn off all the flux in the solder. So, a quick wipe with the iron is all that’s needed to get the solder to solidify on the pads. secondly any solder on the iron tip has no flux left so you should not apply the soldering iron to a pad that has no solder paste on it, otherwise the solder that will stick to the pad from the iron tip will have zero flux.

mwl168, I had a look at the bom for the gr78xx and tried to build a mouser project from it. However mouser says the kst42mtf is obsolete and out of stock. Also C3 C7 in the bom is the same as c1 and c2 but is on a seperate line... the gr79xx also uses the obsolete kst42mtf. the gr79xx also has two entries for the same cap, c1,c4 and the line c7. Ok looks like farnell still has some stock remaining....

Mini T2 Switch off thump. I have found that when switching off the mini t2 (golden reference LV and HV supplies) I get a thump on both channels around 3 to 4 seconds after pressing the off switch. I get no thump on switch on using a 45 second delayed HV so I guessed the LV was dropping faster than the HV at switch off. This was confirmed with a multimeter. Compared to the blue hawaii, the mini t2 drains the HV caps at a slower rate but drains the LV caps faster. I solved the switch off thump by replacing the standard 4700uF caps in the Lv supply with 22000uF. This provides significantly longer LV voltage on switch off and stops the thump. I used 30mm diameter Nichicon 22000uF 35V caps which just fit in place of the 50V 4700uF caps. I suspect 15000uF would probably work to. regards and happy mini t2ing