[regal]

dB,

Here is a link to P Milltets ultrasonic oscillator DHT filament supply. The parts list is so vague it would take a damn electrical engineer to figure it out

Note that he says he encourages someone to polish up the design and use it for DIY.

HF filament supply

 

[nikongod]

Quote:

Originally Posted by FrankCooter /img/forum/go_quote.gif All DHT's require seperate filament supplies. In most cases, DC is reuired. All the usual methods (battery, brute force, volt. reg., current reg., etc.) have their own problems. Probably the best, but the least used, is ultrasonic AC. Something like a Colpitts oscillator running at 50khz into a small broadband amp might be a possibility. You could probably even make money selling it in board or kit form. I'd buy it!

Indeed.
Since no actual tube amp design has come from this thread perhaps the goal could shift to constructing a PCB and running a group buy for an HFAC filament supply.
Someone should ask any of the people who have posted their designes for this for permission to run the boards in limited QTY's and get the ball rolling!

 

[dBs]

Well there certainly seems to be enough interest here to warrant a further look. I'll see if I can decipher this high frequency oscillator and do further research into DHT (I did a bit already but I will do more). As the 300B is a biiit out of the price bracket I was aiming for I will have to find some common DHT heater specs so I could tailor it to that/those tubes.

 

[dBs]

Researched AC heaters the last couple days (been a busy couple of days). There is a bit more that I would like to look into though. I will be trying to look at DC heaters the next couple of days. At this point I would be more inclined to go with a DC though what I have read about them usually indicates that they result in underwhelming performance. Surely there must be a good way to make one =D I have definitely found some good links though regarding AC heaters.

 

[regal]

Quote:

Originally Posted by dBs /img/forum/go_quote.gif Researched AC heaters the last couple days (been a busy couple of days). There is a bit more that I would like to look into though. I will be trying to look at DC heaters the next couple of days. At this point I would be more inclined to go with a DC though what I have read about them usually indicates that they result in underwhelming performance. Surely there must be a good way to make one =D I have definitely found some good links though regarding AC heaters.

The only way to make a good DC heater for DHT is for it to have a high impedance to AC (audio signal) that way it doesn't pickup noise, there are still drawbacks. See Rod Coleman's design on DIYAudio.

 

[dBs]

I've been really busy but researching the vagaries of the DHT PSU. At this point, I am getting the suspicion that taking on something like this might be a bit more than I could chew as a first project. The complexity significantly goes up not just in the amp in general but also in just that PSU set of components. While I am an ambitious fellow, I suspect this might be a bit outside my abilities at this point.

I am still strongly leaning to the Mu follower with //feed. I have some ideas in mind on how to make it happen already and will be looking to start breaking out the math maybe this weekend if an LA trip isn't on tap.

 

[dBs]

I'm starting to hone in on the tubes I am thinking of using. Right now I am steering for dual triodes and pentodes to keep costs down. I can share tubes for each channel or...

I found a gem piece of tube advice on TubeCad where a person pointed out a nice cheap tube that seems to have gone mostly unnoticed; the 6GU7. Aikido All-in-One HPA/LSA & The 6GU7

Looks nicely linear, middle of the road mu, VERY cheap. With a tube that cheap I could make each channels tube unique and just strap the two triode halves together to boost the gain while hopefully not unsettling performance too much.

I am debating a pentode in the upper position of the Mu Follower for the high mu value. The question will be if the added noise is a worthy trade off for the increased performance...as is usually the debate.

I am having a heck of a time trying to find choke components in Electronics Workbench (using version 10). Nothing comes up on index searches. The tube selection is really limited as well. I haven't looked yet, but I suspect that I can find parts that I can add to the programs database or modify the parameters to reflect my tube choice.

I'm sure that I'm the last one to the party on this site but if anyone hasn't found it yet, a great visual "calculator" for tube parameters based on resistor values, capacitors, B+, etc.: Cathode Bypass Capacitor Calculator

 

[fishski13]

i've been looking for a fun tweaker headphone project, 2A3 or 300B. from what i've researched, DC with CCS seems to be the easiest compromise for the filaments. i think head-fier dsavitsk has built a quiet DHT headphone amp with a simple LM317 configured as a CCS, but his website has been down for the last few days.

would a toroid be better than an EI?

as far as schematics go, i've really only put some time in looking at the many Loftin-White 2A3 variants.

 

[dsavitsk]

Quote:

Originally Posted by fishski13 /img/forum/go_quote.gif from what i've researched, DC with CCS seems to be the easiest compromise for the filaments. i think head-fier dsavitsk has built a quiet DHT headphone amp with a simple LM317 configured as a CCS, but his website has been down for the last few days.

Website should be back up soon ... I hope.

I use a LM317 regulator followed by a CCS and it is very very quiet. I tried LM317 based CCSes, but found that they never sounded quite right and were likely oscillating. I switched to a basic mosfet + BJT and it works great. This works fine for low current tubes, but once you are doing high current it is a little more work.

Quote:

Originally Posted by fishski13 /img/forum/go_quote.gif would a toroid be better than an EI?

An EI is generally better for power supply duty as EI power transformers don't pass high frequencies and toroids do. Toroids are used because they are more efficient, lighter, and cheaper and they don't radiate as much meaning they can be closer to other stuff. But, if you have the space, and EI is generally preferable.

 

[dsavitsk]

Quote:

Originally Posted by dBs /img/forum/go_quote.gif I found a gem piece of tube advice on TubeCad where a person pointed out a nice cheap tube that seems to have gone mostly unnoticed; the 6GU7.

Worth playing with, perhaps, but the curves I found don't inspire confidence.

 

[dBs]

Ewww, yea that certainly isn't the curves that were shown over at TubeCAD, which looked much more promising. Hmmm. I can't read the values too well in the graph you posted so I can't tell if your graph represents the same ranges as the one on TubeCAD.

I would imagine that such plate characteristic graph inconsistencies aren't uncommon in the world of tubes given their construction and manufacturing processes of the time.

 

[dsavitsk]

Quote:

Originally Posted by dBs /img/forum/go_quote.gif I would imagine that such plate characteristic graph inconsistencies aren't uncommon in the world of tubes given their construction and manufacturing processes of the time.

More relevant probably is that the curves were drawn by hand.

The ones I posted are from the RCA datasheet. The ones on Tubecad are from the GE sheet.

TDSL Tube data [6GU7]

But, even looking at those closely, I still don't think they look great. Certainly not as good as a 6SN7. If you want an oddball with similar mu and Gm but with fantastic linearity, look at the 2c22.

TDSL Tube data [2C22]

 

[fishski13]

dsavitsk,
thanks. this will be my first "design". i'll no doubt have more questions over the next few months for you and others much more knowledgeable than me.

i have TurboCAD, but what electronics schematic software would people recommend?

 

[dBs]

Quote:

Originally Posted by dsavitsk /img/forum/go_quote.gif More relevant probably is that the curves were drawn by hand. The ones I posted are from the RCA datasheet. The ones on Tubecad are from the GE sheet. TDSL Tube data [6GU7] But, even looking at those closely, I still don't think they look great. Certainly not as good as a 6SN7. If you want an oddball with similar mu and Gm but with fantastic linearity, look at the 2c22. TDSL Tube data [2C22]

Wow, good call on the 2C22. Did some brief research today and couldn't find anyone who didn't like it. The price is certainly right with them as well. I definitely think you found a winner there, thank you =D

 

[dBs]

At this point I am thinking of implementing the //feed using a constant current source instead of a choke. This makes me wonder if the mu follower is pointless, or rather the upper tube is pointless. The idea is that the mu follower replicates (while not perfectly) a ccs. This isn't entirely true when a load is connected, never the less, I suspect that the ccs would make the upper tube portion of the mu follower pointless.

I am also concerned that if the headphones/load is removed, then the top tube is acting as a ccs, but in series with an actual ccs. I know that that is generally a no-no in design to put two current sources in series with one another. I haven't found any mu followers with a ccs after B+ and I suspect that these are the reasons.

I suspect that another reason a mu follower is rarely mixed with a //feed is because the //feed capacitor would have to be a fairly large value due to the low output impedance of the mu follower architecture. I am wondering if my thinking is correct on these points, or if I am thinking about it all wrong.

Otherwise, I found a bit of time today to look at the 2C22 curves and throw some numbers together. I like the values that I am getting. It feels good to be doing math again, lol. My native language.