New build advice

[pdf64]

With a Pi filter, low resistance would probably be preferred, eg the Stancor.
But as you want to lose some voltage with a choke input filter, dropping voltage across the choke may not be a drawback, so the Drake may work out best.
Theoretically the choke input filter output should be very well regulated, thereby any worsening of that regulation due to dc loses may just put you back where you would have been with a regular cap input filter.

I suggest you experiment before committing to which choke you decide to use.
I remember textbooks advising that a low value reservoir cap is used after the rectifier, otherwise the choke might buzz; just big enough to stop the buzz, otherwise it will make a regular cap input Pi filter, eg 1uF.

[Phil_S]

I ran the various chokes through Duncan's PSU. The PSU has a helper for the PT to take the off load voltage and secondary DCR as an inputs, predicting about 405VAC with a load. This seems reasonable to me and suggests about 5.5% drop under load.

The model uses a 5U4GB, a 10uf cap, and a 5K load. Here are the B+ results at the + side of the 10u cap:

Stancor 60Ω 4H 250mA: 363VDC
Drake 445Ω 20H 100mA: 321VDC
Thordarson 124Ω 12H 200mA: 341VDC
MIL/CET 179Ω 8H 150mA: 338VDC
UTC 253Ω 12H 125mA: 333VDC
Thordarson 290Ω 8H 120mA: 331VDC (this is the small one)

I fiddled with the load value some, but had to get down 1K to make a meaningful difference in the result for current at the choke. Voltages didn't change much with respect to reducing the load factor. Current at the choke was generally 70-80mA except for the Stancor at 99mA (no surprise with such a low DCR.)

Curiously (or maybe not) my doodle rule of thumb calculations earlier gave a range of about 325-366, which is consistent with the above results.

I am inclined to build out a breadboard test now and will probably start with the 290Ω Thordarson, which is rated at 120mA. I'd be equally willing to use any of these chokes. I believe PDF64's instinct about the Drake is spot on, but I'm a bit concerned about the 100mA rating because I'm not quite sure what tubes I'll be using. Certainly 100mA will be appropriate for a pair of 6V6, but I think more is needed if I want to use EL84's.

The really good news from the PSU simulations is the voltage drop is close to what I'd hoped for. I believe B+ will actually be a bit lower when put into a real circuit. I think that 5K load is too high and a lower load should pull more current.

It may be a few days or so until I have time to breadboard this.

Thanks to all for the great input. It's been very helpful.

[Meat&Beer]

The one amp I built with choke input is certainly very solid feeling to play, it's very, literal, if that makes sense. It doesn't sag much at all, really. You've got to work for it if you want finesse to come through. That's what I found with mine anyway.

It was also pretty heavy on the low end, my coupling capacitors ended up being .0022's IIRC.

But, for the type of music I predominately play, this was a great fit for me in the end. However, mine is SS rectified, and I believe I saw that you were going with a 5U4GB? Not gonna lie, I love that tube. I'm working on a real shred-tastic amp right now, and it uses a 5U4GB.

On the other side of the spectrum, I just built a super-dee-duper-ly clean amp for a friend recently, and it used a 5U4GB when it was being prototyped. He absolutely loves it! In the end we had to go with a sag resistor, because it's UL capable and wanted a big reservoir cap. Anyway, we thought the 5U4 sounded a tad better in the "alive" department. Just by a nose, though.

So, you might end up kinda splitting the difference with yours as compared to what I found with mine.

Phil, how do you seem to find all the info for the old transformers and chokes you have??

[Phil_S]

M&B: Thanks very much for the notes on choke input amps.

As for info on old iron, as long as it is major brand like Stancor, Merit, Thordarson, Halldorson, UTC, and a few others, there are catalogs available on-line that show specs and they are very easy to find. Heathkit items are often made by a major winder like Stancor, so the trick is finding a good cross reference. A few years ago, someone with rights to Heathkit intellectual property asserted its copyright and made the documentation scarce, which I think is a huge bummer. One well known repository is the Bunker of Doom...look it up. There are other sources, too, just use Google. Info on old iron from Motorola and Woodward-Schumacher is almost impossible to find unless you know what amp it came from. For that you need a lucky hit on the Sams Photofact series -- this is really a needle in a haystack.

Every now and then, if you get lucky, you'll run across a good cross reference sheet for one of the major brands that might tell you what their replacement model is for a given item and then you can just assume the original is the same or very close. Sometimes these are part of a catalog.

For old amps, if you are patient, many local libraries have access to Sams Photofacts, so you don't have to buy them. They might even PDF it to you or you can copy it for 10¢ a page. Don't buy these for $5-$10 each. Transformer specs will be clearly shown in the BOM section.

[Firestorm]

Choke inputs are "pre-sagged" if that makes sense. I'm only familiar with their effect on tube rectifiers, not SS. Bear in mind the voltage drop numbers are typically quoted at full load and half load, so VDC will not drop as much as you think until you're cranked. You can always put series resistance at the front to drop DC if you need to. This is always successful in SE and other Class A, not so much in AB. With 100mA to work with, you could certainly run a pair of 6V6 in any class, but you won't have enough current to push a pair of 5881s or 6l6s or EL34s. EL84s you could do.

So do you want PP? Or SE? An EL34 SE is magical. Plexi sound at 12ish watts. Need the right OT, tho.

[Phil_S]

Test results! I tested the 290Ω Thordarson with a 10uF cap on the downstream side of the choke. Line voltage might be low or maybe it's that bit of load...the secondary was at 846VAC. The naked measure made quite a while back was 855VAC. Meh, that's not enough to consider.

Two 1n4007 diodes were the FW rectifier.

There were 4 dummy circuit loads between the junction of C1 and the choke: 9.6K, 6.2K, 4.5K, and 3.5K. I did not go lower as I thought there was a bit of smell at 3.5K, which I suspect was the single 25W resistor, not the iron, but not taking chances.

mA calculations were made based on the voltage drop across the 290Ω choke. Ending B+ was read between C1 and ground (end side of the dummy load.

Results:
6.9K load: B+359 @ 80mA
6.5K load: B+355 @ 88mA
4.5K load: B+347 @ 103mA
3.5K load: B+337 @ 120mA.

The choke is rated at 120mA. That is the other reason for quitting. I could have set the dummy load as low as 1K or put the various resistors in parallel for even less, but that didn't seem prudent. I believe I have the information I was seeking.

If we discount the B+ reading 11% for the assumed differential between the 1n4007's and the 5U4GB, B+ will range between about 300 - 320 VDC.

There is nothing like empirical data. I believe I'm going to end up in a very happy place with this as a tube rectified power supply. Curiously, the earliest estimate was around 300V!

If you exhibit enough curiosity, I'm willing to put one more choke in the test rig and post results, but I don't really feel urge. Go ahead and egg me on if want to.

My next choice would be the 253Ω UTC. I'm a bit reticent to use the lower Ω chokes without a higher wattage test load. Right now I'm using 25W resistors in series. Maybe I don't understand this. I'm thinking no matter how many 25W resistors I string in series, all I have is 25W rating. I need to put 2 in parallel to get 50W rating. Is this right?

I hope this is useful stuff for all of us. Once again, thanks to all who helped. You guys are great!

Firestorm: thanks for your thoughts. I will try to digest them. I expect this will be a PP amp. My preference all along has been for EL84 or 6V6. 300 B+ puts 6AQ5 in play, too. I don't think I have an output transformer for a single EL34. I'll be looking at OT's on the bone pile now that I have a clear idea what B+ to expect.

When I pick a circuit, I'll start a new tread.

[JazzGuitarGimp]

You need to calculate the dissipated power in the load on a per-resistor basis. If you have two 25W resistors of the same value in series, then no additional calculation necessary - you have a 50W load. If the value of the two resistors differ, then for each resistor: square the voltage across that resistor and divide by the resistors' value. Note the voltage across each resistor will be different if different-value resistors are used.

[Phil_S]

Ok then, thanks for clearing that up. There are 4 x 25W in series. By using a jumper, I am able to adjust the load. At 3.5K there is only one of them @ 25W. At 6.9K it is all 4, 100W. No wonder I smelled something when I got to 3.5K. I can create a similar load by simply adjusting where I put the jumper and I can avoid using a single resistor.

[Phil_S]

Firestorm, I've got the MPS OT10-SE, 5K or 7K to 4/8/16. This is installed on what I consider to be an experimental failed build, so I'm ready to pull it. Do you think this might be appropriate for the SE plexi?

According to what I see, at Va=250, Ia = 105, Ig2 = 15-29, EL34 needs Zout = 2K. I can get 2.5K into 2/4/8. With choke input, I should have the mA. Or should I really be looking at something else for an OT?

I've got a Merit A-2904 "universal" rated for 24W that will let me get close to 8K: 8/16, and a pull from a Pilot that is about 9K:4/8/16. I've also got a few with primary 6.6K, 5K, and 4K -ish. I am thinking I can give a good look at the Merit and the Pilot. There is also a Stancor A3304 with 7K/10K pirmary. I think the 6.6K, 5K and 4K are ruled out as doubling those is a waste when I've already got others in that range.

You can see, I'm wanting this to be a bone pile build. I haven't ruled out buying the OT, though. What would you be looking at for a SE EL34 OT?

[Firestorm]

I'm running mine at 380-390 VDC into 5K. Really special if paralled into 2k5, but you don't have the current.

[Firestorm]

The issue with SE transformers is "bias current" which is how much standing current the core can take before it saturates. No fire and smoke, just loss of bandwidth. I paralled two Hammond 125 ESEs. Even that drops lowend response, but SE is known for that.

[Phil_S]

How much current capacity do you think I'd need for the SE EL34? I'm looking at data sheets and they are not so easy for me to interpret. Based on what I see, it looks like it's somewhere between 120mA and 150mA, all in with preamp tubes. I can use a choke rated for more mA (and run the B+ testing). According to the Hammond rectifier document we've all seen, current out for choke input is 1.54x. I understand this to mean I can have as much as 154mA available from a 100mA secondary. If I understand this correctly, I should re-run the tests using the 179Ω MIL/CET 8H 150mA choke. Thoughts? Am I headed in the right direction? Thanks.

[Firestorm]

What B+? Higher VDC the tube limits its own current (with an appropriate Rk). You just don't want to run the PT at the edge. In P-P Class AB you can stretch things because the current is only feeding one side at a time (more or less) and the caps supply it, they just have to be topped off. Class A is full on all the time.

[Phil_S]

What B+? Higher VDC the tube limits its own current (with an appropriate Rk). You just don't want to run the PT at the edge. In P-P Class AB you can stretch things because the current is only feeding one side at a time (more or less) and the caps supply it, they just have to be topped off. Class A is full on all the time.

I appreciate that you are saying this.

Yes, I understand, reasonably well, there is an inverse relationship between current and voltage and the thing about Class A being all on all the time. Let's say the amp is done with ss rectifier and the 179Ω 150mA 8H choke. I ran the test. Interpolating the data, the graph tells me I'll have 150mA at B+=325. The load line seems to be at around 2.5K. I suppose I should test it at that load and see what it is for real.

I hope to heck my test rig is made correctly. It is simply a ss rectifier, choke, cap, and dummy load (bunch of 25W resistors) . It's on a block of wood. Should I post a schematic of it?

That's it for tonight. I'll see if I can draw the schematic in the morning. To be continued....

[Firestorm]

My SEs are SS rectified 50uF cap followed by 100R followed by 50uF (that's to tame the 120Hz). 1Kz to the screen supply, that's 32uF. Tap the screen supply and divide down to 75ish VDC to feed the heater center tap (quiets the heater, but mostly because I have a cathode follower in the pre). The pre is straight Fender 1st stage, DC coupled cathode follower (hence the elevated heater tap). This feeds the effects loop and the second stage, but the loop is virtual cathode and designed for small gain on the dry side. Final dry gain is paralled triodes, slightly cold. I'm selling this amp. Don't you dare...

But the gain is interesting. It's very high-frequency. Not sure where it came from. Clean, it was amazing. But with gain, it was impossibly hard to play with my limited skills. Knocked it off at the first pre and that was it.

SE amps are underappreciated. Wildly inefficient. But this one gives you dirty at 6 to 30 watts. And clean from 22 W on down.