This project has been discontinued, apologies for any inconvenience.
Last edited:
Here's an update as to where I am at right now.
I've laid the components out on a board matching the size of the original board, which doesn't leave room for the snubber circuit. I am debating if it would be better to slightly increase the board size, for that, or swap the resistor footprints to vertical mounting and see if more space can be squeezed out of the existing size?
KiCad 6 didn't have a built in footprint for a 20mm diameter 10mm pitch capacitor, the one I found online didn't have a nice filled in spot for the negative side of the cap. As mentioned I am a novice at this, so I just drew some evenly spaced lines, but would appreciate if anyone has a source for a more proper looking cap footprint, or maybe there's some sort of repository of better footprints than what's built in?
My solder pads might not be the best choice either, I picked OD 3.9mm ones listed as "1.5sqmm_1x01_D1.7mm_OD3.9mm", honestly I am not sure what's really appropriate here.
If anyone has any thoughts on component layout I am all ears, so far I'm just matching the original.
I've laid the components out on a board matching the size of the original board, which doesn't leave room for the snubber circuit. I am debating if it would be better to slightly increase the board size, for that, or swap the resistor footprints to vertical mounting and see if more space can be squeezed out of the existing size?
KiCad 6 didn't have a built in footprint for a 20mm diameter 10mm pitch capacitor, the one I found online didn't have a nice filled in spot for the negative side of the cap. As mentioned I am a novice at this, so I just drew some evenly spaced lines, but would appreciate if anyone has a source for a more proper looking cap footprint, or maybe there's some sort of repository of better footprints than what's built in?
My solder pads might not be the best choice either, I picked OD 3.9mm ones listed as "1.5sqmm_1x01_D1.7mm_OD3.9mm", honestly I am not sure what's really appropriate here.
If anyone has any thoughts on component layout I am all ears, so far I'm just matching the original.
Last edited:
Tentatively thinking about this for a grounding layout. I'll get back to it tomorrow.
The left side is a ground plane on the top and bottom stitched with ~5-600 some via holes, elevated by C26, R11, and R12 to 1/4 B+ voltage. On the right side is a top and bottom ground plane stitched together with anouther ~5-600 via holes.
The left side is a ground plane on the top and bottom stitched with ~5-600 some via holes, elevated by C26, R11, and R12 to 1/4 B+ voltage. On the right side is a top and bottom ground plane stitched together with anouther ~5-600 via holes.
I just generated two sets of gerbers, one with vias, one without. I then loaded both into jlcpcb, they are the same price for 5 boards, 20 boards, etc, options selected were default of 1.6mm thick, 1oz copper.
That said, deleting the vias is a trivial task so I will try to provide gerbers in both formats in case other PCB houses are going to charge extra people on vias.
I believe if you choose the option “tented vias” they are covered with solder mask which eliminates the chance of solder bridges when soldering components close to an exposed via creating a short to gnd.
Looks like KiCad by default has tented vias, I created another gerbers with no tented vias, still the same price on JLCPCB so perhaps they are just kind with pricing over there.
Good to know, well regardless in the interest of making sure there are options for everyone without people needing to go through KiCad, I'll try to remember to generate gerbers with vias, un-tented vias, and no vias. Tomorrow I'll get cracking on making sure the silk screening looks good, and we'll take it from there.
I have this board from ebay and it works very well (though I no longer like the sound of three terminal regs). I read in Blencow’s book that Q1 being a BJT is susceptible to shorts. I found this out first hand. He recommends using an FET in its place. I did this mod on my board and it works perfectly. You do need to change the zener D1 to leave 6V over the regulator plus about 4V Vgs. So a 9-12V zener works. I’d recommend this change, but you should test it yourself before committing to a board.
Hi bloqhed, here are some comments on your first post here in this thread.
HV delay is desirable - any thoughts on how to implement this yet?
For the trimpots in both HV and LV regulators, these are shown wired as variable resistors. Most variable reg circuits will have the trimpot wired as a potentiometer, with one end of the pot and the wiper connected to GND. This way if the wiper goes open circuit, the output/Adj is still referenced to GND. With the way it is currently drawn, if the wiper goes OC then the output will rise to the input voltage with damage likely to the reg chip and connected circuit (tube filaments for instance). Also normal convention is to have a fixed resistor between out and adj pins with the trimpot from adj to GND.
For the GND loop breaker I would just use a CL60 NTC thermistor as is popular in the Pass class A amps - this will save on components and pcb space.
WIMA caps are good as are the Nichicons you specify, but a quick look has most of these out of stock at the moment, so alternatives (different voltages) will need to be sourced that fit the pcb spacing.
I am not keen on vertically mounted resistors - heat rises and the high point of the vertically mounted resistor will run hotter than the lower end. Horizontally mounted resistors with 3W types spaced off the pcb will at least be evenly heated across the resistor body.
The HV pass transistor (BUT11A) is obsolete - use either MJF18004G (450V) or KSC5502TU (600V), they have same pin out. Or, as another member has suggested a suitably rated TO220 Mosfet.
The final PCB size is not an issue for me.
Tented vias are good and should not add to the pcb cost at JLCPCB. Definitely need 2 ounce copper especially for the tube filament currents that might be required. Might be wise to increase the track width on the filament supply.
Metal Oxide 3W resistors in BOM are only available in min order qty of 100. Use 3 Watt Metal Film resistors rated at 750V like the Vishay PR03 types at Mouser - never had a problem with them in HV tube circuits.
You are doing a great job so far, and we will appreciate your efforts for sure.
Cheers.
Gary.
HV delay is desirable - any thoughts on how to implement this yet?
For the trimpots in both HV and LV regulators, these are shown wired as variable resistors. Most variable reg circuits will have the trimpot wired as a potentiometer, with one end of the pot and the wiper connected to GND. This way if the wiper goes open circuit, the output/Adj is still referenced to GND. With the way it is currently drawn, if the wiper goes OC then the output will rise to the input voltage with damage likely to the reg chip and connected circuit (tube filaments for instance). Also normal convention is to have a fixed resistor between out and adj pins with the trimpot from adj to GND.
For the GND loop breaker I would just use a CL60 NTC thermistor as is popular in the Pass class A amps - this will save on components and pcb space.
WIMA caps are good as are the Nichicons you specify, but a quick look has most of these out of stock at the moment, so alternatives (different voltages) will need to be sourced that fit the pcb spacing.
I am not keen on vertically mounted resistors - heat rises and the high point of the vertically mounted resistor will run hotter than the lower end. Horizontally mounted resistors with 3W types spaced off the pcb will at least be evenly heated across the resistor body.
The HV pass transistor (BUT11A) is obsolete - use either MJF18004G (450V) or KSC5502TU (600V), they have same pin out. Or, as another member has suggested a suitably rated TO220 Mosfet.
The final PCB size is not an issue for me.
Tented vias are good and should not add to the pcb cost at JLCPCB. Definitely need 2 ounce copper especially for the tube filament currents that might be required. Might be wise to increase the track width on the filament supply.
Metal Oxide 3W resistors in BOM are only available in min order qty of 100. Use 3 Watt Metal Film resistors rated at 750V like the Vishay PR03 types at Mouser - never had a problem with them in HV tube circuits.
You are doing a great job so far, and we will appreciate your efforts for sure.
Cheers.
Gary.
Thank you! I am going to look into the FET, I imagine I would need to draw this circuit up in LTSpice and fiddle around with swapping out the BJT for a MOSFET? Do you have a preferred replacement to 3 terminal regulators for sound?
Cheers! I appreciate the kind words and insight into the project.
HV delay I am still looking into, I think it will definitely be the largest hurdle here.
- Jan Didden has his circuit that looks promising, same one sold on the diyaudio store and it looks like the audioxpress article includes the schematic
- ESP has a circuit but it looks like it needs a dedicated 6.3vac winding which is less ideal than Jan's; High-Voltage Time Delay
I'm hoping I follow your comment on the trimpots correctly, is the fix as simple as just terminating pin 1 on both of them to the ground like in this picture?
I'll get the loop breaker circuit swapped out for a CL-60 NTC, and perhaps a 22nf Y2 cap to keep HF coupled, Rod Elliot states "The capacitor in parallel ensures that the electronics are connected to the chassis for radio frequency signals, and helps to prevent radio frequency interference."
For the electrolytics I'll do some browsing on mouser and digikey, but tentatively I am assuming I'll swap the footprint over to 22mm diameter/10mm pin pitch for more versatility.
I'll look into MJF18004G as it has a higher power dissipation over the KSC5502TU, then start looking into the idea of a suitably rated FET, but I assume this means I'll need to zip over to LTSpice and run some tests? I see BUT11A's in plenty of stock at mouser (link), and has good power dissipation, so not sure if we still want to consider the MJF18004G, especially if tentatively we're looking at switching over to a MOSFET. At a quick glance, STF8NM50N seems like it could be a viable option for a MOSFET.
I am going to add a new column into the original BOM with what we'll call "updated parts" for now. Quick glance the PR03 resistors look to be the same size as the 3w metal oxides. A good point on vertical mounted resistors and heat, I didn't think about that. Heat is why I added 3 holes beneath the large 5W R4.
I have started leaning towards zener followers, but that’s a different topic. What you have here is very much needed and worthwhile!
I changed to the STF3LN80K5 part, just coz I had them. 800V and plastic package. Win, win.
https://www.st.com/resource/en/datasheet/stf3ln80k5.pdf
I changed the zener to 12V and I can’t remember if I changed the 22k resistor. It needs to source current for the zener. That’s all it does.
Oh, you should also add a resistor from MOSFET gate to prevent oscillation.
I changed to the STF3LN80K5 part, just coz I had them. 800V and plastic package. Win, win.
https://www.st.com/resource/en/datasheet/stf3ln80k5.pdf
I changed the zener to 12V and I can’t remember if I changed the 22k resistor. It needs to source current for the zener. That’s all it does.
Oh, you should also add a resistor from MOSFET gate to prevent oscillation.
Interesting you use 0.1uF for C13. Other maida regulators I've researched on the web either used no cap or 1uF. The 1uF then has to discharge somewhere on a short - typically through the diode around the LM317/LDO with spike of current.
That STF3LN80K5 SOA seems small at higher voltages (ie startup) did you have a problem with it? I've just bought a set of STW9NK95Z (TomChr's favourite) with a spare but would have preferred an updated part. I did find a few but they're 5-10x the price. So I'm on the hunt for a replacement in future.
That STF3LN80K5 SOA seems small at higher voltages (ie startup) did you have a problem with it? I've just bought a set of STW9NK95Z (TomChr's favourite) with a spare but would have preferred an updated part. I did find a few but they're 5-10x the price. So I'm on the hunt for a replacement in future.
Last edited:
I use it with a tube rectifier (type 80, which is not a slow start) and have not had problems so far. There are many other candidates. I used what I had on hand. I do like the plastic package though coz you can bolt it directly to the heatsink.
Thanks for all this, I'll read up on how to calculate gate resistor values.
To be fair, I'm a complete novice and learning as I go. 🙂 In the first post the bottom schematic in white is the original one found on ebay that is no longer available for purchase, the top schematic from KiCad is mine. At this point I'm just working at recreating the board that was previously sold on ebay with some free for everyone gerbers, and while we are at it, whatever improvements we can come up with to make this more well rounded. If there's a benefit to 1uF over 0.1uF for that capacitor I am all for it, but my lack of knowledge here shines through.
Last edited:
I'm not really either. I have my RCRC linear supplies running so the next step is to reduce the ripple further. I have the pieces for my regulators ordered - although mine will not be as neat as your PCB.
Thanks bloqhed, for the HV delay using ESP circuit, you can use the 6.3VAC supply that will feed the filament regulator. he also has a delay circuit using a 555 timer IC that mutes the audio output at the amp. If using the HV delay than the main point is to use a suitably rated relay with contacts rated for 250VAC with both double pole contacts wired in series to switch the HV AC supply to the regulator. I have not seen Jan's circuit yet - but will have a look.
Yeh the trimpot is as I said, but normal practice is to have a fixed resistor between Vout and ADJ pin and the trimpot in the lower position with the wiper (pin 2) and pin 1 to GND. Or to be more safe a fixed resistor in place of the pot calculated for the exact filament voltage. In your circuit with 100 ohm resistor between Vout and ADJ, the bottom resistor would be 408 ohms based on the IC ref volts as being 1.240VDC, for a 6.3VDC supply and 916ohms for a 12.6VDC supply. Use the nearest 1% value = 407ohms and 910ohms.
All good with your comments on the CL-60 NTC.
Yeh as long as you use a common sized electro cap footprint and lead spacing, then some shopping around might have to be done on the caps.
When I looked at Mouser for the BUT11A it said it was obsolete, but your link shows in stock, the FET might be a better option here though.
The PR03's are good. What have you picked for the 5W resistor R4 - it is wrong on the BOM. Maybe this one: 279-ROX5SSJ1K0 at Mouser.
Keep up the good work, much appreciated.
Yeh the trimpot is as I said, but normal practice is to have a fixed resistor between Vout and ADJ pin and the trimpot in the lower position with the wiper (pin 2) and pin 1 to GND. Or to be more safe a fixed resistor in place of the pot calculated for the exact filament voltage. In your circuit with 100 ohm resistor between Vout and ADJ, the bottom resistor would be 408 ohms based on the IC ref volts as being 1.240VDC, for a 6.3VDC supply and 916ohms for a 12.6VDC supply. Use the nearest 1% value = 407ohms and 910ohms.
All good with your comments on the CL-60 NTC.
Yeh as long as you use a common sized electro cap footprint and lead spacing, then some shopping around might have to be done on the caps.
When I looked at Mouser for the BUT11A it said it was obsolete, but your link shows in stock, the FET might be a better option here though.
The PR03's are good. What have you picked for the 5W resistor R4 - it is wrong on the BOM. Maybe this one: 279-ROX5SSJ1K0 at Mouser.
Keep up the good work, much appreciated.
Last edited: