The Donut Princess Build

[mirage_indigo]

So I should have left myself way more space on the power supply board for the filament and bias supplies (also the bridge rectifier). I got it in there, but it's way too close together.

IMG_20200806_231959.jpg

[john_a]

Great stuff in this build! Excuse my ignorance as I've just joined this forum yesterday, and don't have any design OR build experience, but would it be possible to lower the big filter cap by sinking it into a wide slot cut in the board? If that cap is 54mm dia, a one-inch width slot could drop it about the same as changing standoffs from 1/4" to 1/8".

Maybe that's not feasible, but it was the first thing I thought of upon seeing how far it had to move for clearance.

/John

[pompeiisneaks]

I think the likely problem with that approach is that for very large caps, in an environment with vibration, you need to attach them to something (the board) to keep the leads from breaking due to strain. I'd guess it would still be feasible but you'd need to design some means of still holding the caps to that same board. Zip ties? glue to the sides with silicone?

~Phil

[john_a]

I think the likely problem with that approach is that for very large caps, in an environment with vibration, you need to attach them to something (the board) to keep the leads from breaking due to strain.
~Phil

I was thinking of it as the cap's still on the board, but recessed a little into a rectangular slot cut out of that board under where the cap sits now. You'd end up supporting it on the two edges of that slot, instead of a single line of contact as it sits now on the flat surface. So if it did not weaken the board too much, it actually would secure the big cap better vs the single line of contact it normally has.
With 54mm dia cap sitting in a 27mm wide slot, it drops down 3.6mm. That's the "segment of a circle" calculation that led me to suggest 1" width slot would provide about the same clearance as using 1/8" standoffs instead of 1/4". And because that lower segment of the cap would mostly fit in space that was board thickness, you'd retain almost all of the clearance under the board you lost from going to 1/8" standoffs.
/John

[mirage_indigo]

I actually did consider cutting a rectangular slot in the underlying board to lower the main cap. I'm using 1/8" G4, so plenty strong. The main reason I didn't do it was G4 is a PITA to machine for large holes (or slots in this case) even with a router table, and I'd need to figure it out with a captive slot, so no bandsaw. I opted for shorter standoffs for the power supply board (basically two 10-24 hex nuts). It solved enough of my problem to get me under the clearance I needed (although not by much). Use a deeper chassis.

The caps themselves have extra thick leads (big enough they don't fit in the mighty keystone 1509-4) so I've never attached them to anything mechanically beyond the solder joints to the turrets (which go around the turrets). They're actually not that heavy. The mechanical failure point is the swagging of the turret itself.

BTW One huge problem with my layout is it's trying to cram too many things onto some of the turrets. If you use my layout, def might consider reworking the layout of the bias and heater supplies to have less things into the turrets, the man issue I had was my turrets (unlike the eyelerts I've used previously) couldn't handle two diode leads into the hole, so I had to wrap the diodes around the turrets. This was hard to do after I'd already swagged all the turrets way too close together (it looked like enough room on the plan).

Hopefully I should have a bunch more progress this week, been on a socially distant vacation with the kiddos. Next up is double checking the three supplies, wiring the heaters and installing the ground lugs bus points. Also I need to go build a lightbulb limiter before I fire up for the supply (HV, bias, heater) test. Possibly also an amp cradle.

[mirage_indigo]

Also note that the easier solution than cutting out a slot midships in the g4 is to just hang the cap off the side of the board. Might do that next build. Or just parallel a 27uf and a 22uf (or whatever) and be done. The next two builds will be a Sorrado clone and a JM clone, so I'll be using the big can 100uF / 630v:

https://solen.ca/products/capacitors/fa ... e10000alv/

[mirage_indigo]

I got back around to wiring up the heaters this week and then built myself (finally) a lightbulb current limiter.

Good thing I did, the heater regulator circuit does not seem to be correct. I fired up just the PS board and wiring with a 25W bulb. It glowed brightly for a bit while the B+ charged up, then went out (since there's no load). I'm getting about 480ish no load on the B+ rail (which is about what I expected), then -45V no load on the output to the top of the bias pot. The B+ stayed high well after flipping the power, so I really need a bleeder resistor in there. It drops about a 1v/sec when the power is off, so a 100k 3W should do the trick. So mains and bias look fine.

The heater circuit using the lm338 is the more odd bit. I'm using a variant of Merlin's DC heater supply. Basically that rectifier bridge design then run into the reference LM338 circuit from TUT (which lines up pretty well with their application note).

I was at first getting ~5v DC to the tubes when the amp was charging up, but then 0v, so I'm guessing the regulator dropped out. What's weird though, is now I'm getting no AC input from the PT and the lamp on the front panel is out. I'm using the twin 6.3vac windings (hopefully) in phase and in parallel. Might I have shorted a winding someplace? The PT doesn't seem hot at all although I obviously didn't run it for very long once I figured out something was wrong. Is there an easy way to tell if the heater windings are in phase?

Heaters_Zoomed.pdf

[sluckey]

Is there an easy way to tell if the heater windings are in phase?

Yes. If they are in phase you will have 6.3vac across the windings. If they are out of phase you will have zero volts across the windings... and likely to burn up both windings.

Disconnect the two windings from each other and use an ohm meter to check resistance of both windings. If you get a small resistance then check for 6.3vac this assumes you are using 6.3vac secondaries) across each winding. If you have proper voltage, consider yourself lucky. Next, twist ***ONLY***one lead of each winding together and measure the voltage across the other dangling leads. If you read 12vac then stop and reverse one lead of ***ONE*** winding of the leads you had twisted together. Now you should read zero volts between the other dangling leads. If so, then twist the two dangling leads together and check voltage. Should be 6.3, meaning they are now in phase.

[mirage_indigo]

Thanks for your help, Steve. The windings are OK. I had them correct, but I think I was shorting one leg someplace else in the lamp connection. I had also blown the main fuse, possibly from not using slo blow + toroid inrush. I unsoldered my lamp/heater terminal and rechecked, replaced the fuse and I'm back in business. Next build I swear I'm going to wise up and allocate a dedicated terminal to every transformer connection. Hooray for lightbulb limiters.

With no current limiter at all I'm getting about 5VAC outta across the windings, but I think the battery in my multimeter is bad as the value jumps around a lot. Both windings are producing voltage. I was able to verify the phase. With no load and no limiter, B+ is 440v which looks more correct for a 320-0-320, bias still at -52. The capacitor coupled bias is working great so far with no load on it. I need to go rethink how I routed the heater windings to the 100ohm terminals and the lamp on the front panel. The mechanicals are constrained much tighter than I wanted.

[mirage_indigo]

Got some time today to futz with the amp more. Still debugging the main power supply board (mostly unloaded).

The PT (antek donut) is getting super hot even with no load at all OR the light bulb limiter is glowing, so I'm drawing alot of current someplace even with no load. With a 25W limiter I have enough voltage I can debug I think. I strongly suspect I have a short or fault someplace in the heater rectifier circuit.

The main B+ node is right on 437V, with V+C and V+D also at the same (makes sense, there's no load). This seems pretty happy. The supply stays charged even after power off for a long long time, so I suspect this is working fine.

The main bias supply is -52 and very stable. This also stays charged for a while, but since there's a DC path to ground through a 56k, not nearly as long.

With the light bulb limiter out, I'm getting 5.4Vac across the in phase heater windings. The regulator is still putting out an almost perfect 5V. So at 6.3V, the regulator should work. I think there's a short in here someplace which is what's causing the drop. It might be the ground reference for the 100ohm center tap. I based it off this?: http://www.valvewizard.co.uk/heater.html + the KoC circuit which is itself based on the Lm338 application note.

What's odd is when I look around the interwebs for DC filament heater supplies using a lm338, None of them have a center tap for the AC side of the windings and instead ground the lower side of the DC regulator.

Any ideas?

[mirage_indigo]

I figured it out. I had a bad diode in the main B+ bridge on one side. Not sure if it was open or shorted. Basically it was dumping AC straight to ground in half the cycle. This let the B+ node charge up but also put enough load on the OT to saturate the core. Once I removed the secondary from the bridge, the heater windings came back up to 6.3 and no current limiting. So not the heater windings.

I figured out how to jam a new set of diodes in a very small space and it's working now. Unloaded I was getting ridiculous high voltage (600+) but with a 15k across the preamp supply node (drawing ~25mA or so), more like 390 B+ and 188V preamp B+ which is much more in line with what I'd expect, but still low. No light bulb limit at all and no transformer heat. Bias voltage still at -51V. I might drop the 1k "choke resistor" down a bit to get some more preamp voltage.

Does anyone have an integrated part they use for mains rectifier bridge? I have a 300V 1A BR3 package I use for regular (non-tube) stuff, but 'm not sure I really want that in the mains path. For one thing, I'm not sure what peak current is in these things for the mains.

So The power supply works and heaters are wired. I need to adjust the regulater output level set resistors, I'm getting a rock steady 5.0V from a 6.98 VAC rectified input. It was also a perfect 5.0v at 5.3Vac input, so I don't think it's drop out, especially since the heaters are currently unloaded. Anybody use an LM338 for heater windings?

[mirage_indigo]

For an LM338, the output voltage looks like Vout = 1.25v + (1 + R2/R1) + Iadj/R2. My R1 is 100ohm and r2 is 300 ohms. This works out to first order to 1.25 (1 + 3) = 5v with a nominal Iadj current, so working as intended.

Min drop out on the part is listed as 0.3V, which I'm sure is not recommended, but that's what I have since I'm getting about 6.9VAC input. To get my 6V, I need an R1 over R2 ratio of: 6/1.25 -1 => R1/R2 = 3.75.

I've currently got two resistors in there and I'd also like to only change one of them. I could leave in the 100ohm and increase the 300 ohm to 375 ohm and it should work. I need to double check I actually installed the right resistors anyway.

Edit: I ended up with 300 ohms over 1.3k ohms and it's giving an output of ~5.8V which is close enough to 6.3 that I'm going to call it good.

[sluckey]

You could use a pot and adjust the output voltage for exactly like you want. Like this...

[mirage_indigo]

I've made some more progress on the basic wiring:
-> power tubes attached to OT primary.
-> OT secondary to speaker switch with local star ground.
-> grid and screen resistors installed.
-> reverb driver daughter board installed. Had an error on the schematic and the layout, will post correction to first post in the series as v1.4
-> power supply board debugged, all 3 supplies (HV, bias, heater) all looking good. Fusing works! DAMHIKT.
-> all preamp tube sockets wired on socket end.

I'm running the reverb OT right on top on the board inside the chassis, we'll see how it goes. The big yellow cap is a 3uF 600V that I put across the power supply lines of the reverb driver since it's pretty far over on the other side of the chassis and there wasn't a good way to twist the power supply line to that board. The circuit is the Ko'C self-split "standard" driver from TUT2 but with a hammond 125A (which is probably overkill by a wide margin). Based on looking at published 1W firefly style designs for 12A_7 output power amps, I wired the primary for 22.5K impedance into 8ohms (lug 2-> ground, lug 4 -> output jack hot). The plate resistance of the 12At7 is still higher than this, but this seems to be as high as you could go on that part. Should be fine.

PXL_20200925_024850905.jpg

PXL_20200925_024903156.jpg

[mirage_indigo]

Got the main board drilled, staked and about halfway populated. Matches the layout in Post #1 pretty well. I decided to add the turrets for the optional Mosfet Trem oscillator driver. Not sure if I'll need it or not.

Blue Wires are those headed to the front panel controls, purple is the main NFB wire from the speaker out, green are grounds. You can see the paralleled 1.5K / 25uF cathode pairs along the left side. Then a bunch of 100k resistors along the bottom of the board (which is towards the power tubes.). The three holes are for the 3 of the 4 12AX7 preamp tubes.

PXL_20200929_224919937.jpg

PXL_20200930_054059164.jpg