6EW7 Dc coupled SET OPT Headamp

I'd consider cascoding the depletion mode MOSFET (it's cheap, so why not. Of course, you could do better than this simple CCS, but it looks like you're trying to cut costs with this design?), as well as using a series resistor between the CCS and the 6EW7 plate. While this second addition might raise some eyebrows, I think there are two reasons to try this. First, reduce heat and second, to isolate the CCS from the tube. You might google around for a certain Mr. Beck's opinions on CCS capacitance, and unless you happen to have an excellent test rig to verify that this isn't causing problems, why not prevent the issue in the first place? (even despite lower Rp tubes, I think this is a good idea).

For power supply, I think you could do much better, for less money. What about the various incarnations of the Maida regulator? I know a few people around here have some boards I made up, and it does indeed work quite well.

Looks good and I agree DC coupling is the way to go. Only out of curiosity, wonder if a simple resistor sounds better than the CCS. 18K load resistor for 10mA might not be enough given section 1 plate resistance.
35K load resistor, -4V Ig, 5mA Ip.

You can replace the bias resistors with a CCS, and in fact I would recommend doing it for a variety of reasons, but you will need to keep the bypass cap -- otherwise you'll ge no sound. If you use a CCS there, tie the return to B+ rather than ground -- that is, use the cap as an ultrapath cap (and don't listen to Broski's anti-Ultrapath article. He's a smart guy, but the circuit he criticizes in that article isn't an Ultrapath, it's a strawman. It is a terrible article.) This works especially well with a CCS in the cathode as that forces all of the AC current through the cap keeping it totally out of the PS. With a resistor bias you still get some through the last PS cap.

You need grid stoppers on both sections.

One other thing -- no need to worry about making the DC connection after power up. The only reason you would ever need to worry is if the output tube was fixed bias with a grounded cathode and you were worried about runaway. But, with a resistor/ccs bias that won't happen. Also, because the two sections are the same tube, they'll start conducting at the same time, so there is no chance for gross imbalance. Last, these are tubes, not transistors -- they are tough and can take out of spec voltages for a while. You won't hurt them. And, if you do, they cost $4 to replace.


This raises my eyebrows -- the tube itself has plenty of series resistance. Why add more? Sounds like a solution in search of a problem to me. And, IMO, Mr. Beck hasn't done anything very good since Sea Change.


I'd be inclined to agree that any three legged series reg is probably plenty here. No need to make it too hard.

No, it doesn't.

what about using a neon bulb, or a pair of zeners in this case? but, yeah, all good points

People do silly things I guess for various reasons. 5W of heat is not insignificant. Also, it gives an easy way to set the CCS current in situ. Whether or not you think the resistor isolates from CCS capacitance would be the real debate. Mr. Yaniger suggested I tried it, and well, it seemed to work well and "sound" better... Either way, I'd suggest to the OP add another MOSFET to the CCS ...

yay!

I like the ultrapath cap + CCS a little more because with zeners to ground you still complete the current loop through a PS cap, so the choice is zeners + cap, or just cap. But, opinions will clearly differ.

I'll look in to it.

Indeed.

I'm not sure I follow about cascode performance of the 10m45s. Granted, I don't use the 10m45s anymore, as I've started using the 900v part instead, but tracking down the Supertek parts is probably worthwhile, if it will work for you (link or link).

Also, this might prove a useful read:

http://waltjung.org/PDFs/Sources_101_P2.pdf

There is?

The quick and easy way is to just consider it a coupling cap into the impedance of the transformer. The even easier way is to use a 50u motor run cap and not worry about it too much.

Now, notice the tradeoff here -- if you run more current, it requires a bigger gap in the transformer to handle the DC. This lowers the performance of the transformer. Plus, it requires a bigger core which further lowers the performance, and increases the price. Sometimes this is unavoidable, and sometimes it is worth the extra current, but it is worth considering. Might be better just to increase the impedance and use lower current.

On that note, single feed transformer coupled amps are good where you need a large output swing as the transformer can swing higher than B+. Thus their use in power amplifiers. As with anything, there is a tradeoff in that you need a gap to accommodate the DC in the transformer, and gapped transformers are not as high fidelity as interleaved. So, instead, since the amount of swing you need is minimal, you might do better to parafeed the second stage -- still use a 5K:32 OPT, but load the output tube with a CCS, too. I'm not trying to derail your design -- just offering some things to chew on.

You describe a lot of different things as sounding "harsh" and "solid statey". Both of these are almost certainly due to oscillation.

I don't see a schematic, but your amp sounds similar to the Gary Kaufman 6ea7 design. Definitely one of my favorite circuits for a simple two stage transformer output headphone amp. I've built 5 or 6 versions of this circuit with different tubes (including the 71a dht) and they've all worked well. Lots of potential here.

On "first pass" construction I'd keep everything as simple as possible. Use resistive loads and bias schemes. Start with a simple tube rectified unregulated power supply with an LCLC filter using motor-run type oil caps. If something doesn't work, you'll be able to debug it using nothing more than a multimeter. You'll also get a good idea of what the basic circuit sounds like. Add modifications, like ccs plate loads, exotic bias schemes, and B+ regulation one at a time. You may or may not find them to be improvements. For instance, I don't mind a ccs plate load, but I don't like solid-state regulated B+ supplies in tube amps. Pick and choose one by one what's best for you.

Regal, if you go with the ultrapath configuration make sure you use a "decent" poly or oil cap, not the cheapest electrolytic you can find.

I've built this circuit, except that it didn't use an LED to bias the driver. You might also want to consider a monkey or free lunch and using a choke load for the driver. But on a cost/performance basis you're not going to beat the CCS, if cost is a factor. (Good plate chokes are not cheap.)