DIY talk

RAAL forwarded. my question to the engineer of the amp, and this is what he said:

"Yes, after many hours of operation, the PDT1a becomes hot. Output vacuum tube 6N13S/6AS7G/6080 works at almost maximum dissipation and conduct 117mA current thourgh 1K resistor. It is 13.7W dissipation on 1K resistor. I use 25W metal resistor per channel which is attached on both sides of the amplifier. So, two resistors always dissipate 27, 4W. Some of the heat is transferred to the front under the faceplate and the knob gets hot. So, the answer is, yes, it is normal to be hot."
​

Reading this again though, my amp gets super hot right away so I may write them back.

 

I know I’ve read makers say they designed their amp to run transformers and tubes at a particular temp to meet certain properties. Like they discourage additional cooling as it would affect the sound.

Best thing I think is to have a reversible mod to compare what it sounds like before and after additional cooling.

 

If tubes are working at their max, lifespan will be greatly reduced

 

YOWZA. The Bottlehead Crack runs less than one-third of that current, and it gets pretty damn toasty. I can barely imagine that much heat dissipation in a box the size of the PDT1a.

In that context, the volume knob getting that hot doesn't surprise me, and the amp is probably operating as designed. But man, whoever thought that design was acceptable in a consumer product.......

 

Maybe just because it sounds f'ing awesome. I suppose the high current is why it sounds so good with Audeze’s.

They at least need to put a rubber ring around the volume knob or something. I can’t imagine head-fiers dealing with this and not burning themselves.

 

Yeah, I mostly meant in terms of heat dissipation. It definitely looks like it needs more ventilation, if nothing else.

 

yeah I agree, I'm almost kind of worried about lawsuits from people burning themselves on this thing, it gets that hot. I just slightly burned my fingers on the volume knob and I was being careful.

Time to stock up on 6AS7G tubes before people realize how awesome this amp sounds and the prices all go up!

 

+1. Dissipating 27W on a chassis that size is bonkers. Safety first!

I get heatsinks from Heatsink USA. They are unfinished aluminum extrusions. For something easier on the eyes, Aliexpress/eBay has a bunch, but it will be trickier to find one made to the size you need.

Some conservative napkin math says that one of these guys (1"x8", ~2.7C/W) per side should bring the chassis down <55C.

 

@rhythmdevils can you get something like this to measure the actual surface temperature?
https://www.amazon.com/Klein-Tools-IR1-Thermometer-Non-Contact/dp/B0873T6SGR/

 

I'm revisiting my first solid state amp, KG's version of the Singlepower Squarewave. Here's the original schematic that KG traced, then updated in 2013.



It's a FET input buffer, a cascode VAS, and a common emitter (!) output stage with current feedback. The VAS is cascoded. This was in 2008! Poorly executed, perhaps, but way ahead of its time-- it's practically a Firstwatt F5.

KG's take on it used a Vbe multipler to bias the VAS, and uses a servo to eliminate the output caps. I think it runs a bit hotter than the original as well. It sounds great.



Complementary push pull amps aren't usually my jam, but I think the design has potential to make a fantastic transconductance amp à la Bakoon. Here's my take on it, with options to build it as either a conventional current feedback amp, or as a "current output" amp without global NFB by simply lifting the feedback resistor.



Some of the big changes:
- Input FET and CCS are cascoded.
- The VAS and output stage are also cascoded, both with Hawksford cascodes. This gives significantly better linearity before adding feedback. I suspect cascode output stages do better with back EMF, as I've found they yield immense improvements to sound.
- Simpler servo
- Fancier Vbe multiplier
- A buffer between the VAS and the output stage. Doug Self took some pretty compelling measurements suggesting why this might be a good idea, and Spice agrees.

I'm hoping to order boards in the next few weeks, and build both variations to compare. Suggestions are welcome, I'd like to make the gerbers available to the community as for the phono project.

 

BHC drove up prices on the goot 6as7g tubes a long time ago...

I may be selling off some of my toobz in another week or two... and will probably have 1-2 tungsol/chatham 6as7g's in the list...

 

Watching with interest.

What sort of bias/power levels will this sit at? Just thinking onboard versus offboard heatsinks, that sort of thing.

Also wondering about availability of the sand. Nothing too exotic or rare?

 

Last time I looked, there's still a lot of solid workhorse NOS options out there like the RCA JAN 6AS7. I quite like the new production Winged C as well. A lot of Crack folk started chasing after rare 6080 variants, not quite sure I understand the appeal.

My favorite in the Crack is the 5998, but at the price they demand there's no way I would want to burn through them fast at 117 mA, no matter how good the amp sounds....

 

A DIY transconductance headphone amp is something that’s been on my radar for a very long time. Ever since I read what Nelson Pass said about his F1/F2 amp, I thought it was perfect for headphones as they are single full-range drivers.

I thought about trying make a headphone version of the F1, but that’s a bit beyond my engineering talent. I’m very interested to see this project.

 

RCA's are also good and I have a couple of them that will probably go as well... I just like the Tungsol better .

There were enough NOS winged C around that I never listened to a new production... I loved the bass on them, but I didn't think they were all that balanced. And I suspect those are going to be hard to get going forward.

I've never heard the 5998... decided they were too much compared to what else was out there... As for the 6080... in the crack I like almost all 6as7g's I've heard better than the 6080. Exceptions would be the tungsol 6080 and the tungsol 7236 (which I guess isn't really a 6080)

 

Thanks for the interest.

I'm looking at 250mA with +/- 12V rails, maybe +/-15V. This works out to 7.5W of dissipation (or about half a 1k resistor in a PDT-1a ) per channel. The boards are designed to comfortably fit on a 2U heatsink, but won't include a power supply.



The servo will run off of the same rails as the amp, so this will ultimately limit the voltage swing. No reason you couldn't run an amp per channel tough. (I may try this for speakers.)

Also, SE only. No good way to run this design balanced, short of using input and ouput transformers.

Nothing exotic, no. Some of the JFETs are SMD, since there are more compelling options than with through hole. Possibly Toshiba's 2SK879 or TI's fancy new JFE150. The trick is that the FET's Vgs needs to be >0.7V to bias the subsequent stage, which means a lot of the high transconductance FETs like the 2SK3557 or CPH3910 won't work. Intersted if you'd know any good candidates!

Edit: the JFE150 looks absolutely miserable to solder.

The TO220 BJTs are 2SC6144SG and 2SA2222SG. I'm not sure why we don't see them more often in audio-- they have exceptionally high gain for power devices, and have really linear hfe/Ic curves despite being advertised for switching.

Definitely! There are some awful sounding transconductance amps (I've built a few ), but when carefully designed they sound very different, but not wrong.

 

I came across this piece on coil winding. Might be of interests to some of you

https://hackaday.com/2022/06/19/tips-for-winding-durable-coils-with-nice-flat-sides/

 

@rhythmdevils
You could try getting a fan and blowing air on your hot amp to cool it off. Moving air will pull the heat out.
Not optimal without heatsinking, but easy to try if you have a fan around somewhere.
See if it cools off, and then if the sound changes.

Randu

 

RAAL wants me to send the amp back to them to check out since I bough it used. It will be hard to go without but I'll report back what they say when they get it.

 

I’m getting some inspiration from here horns



I always thought front horns needed a lathe to make a smooth surface. However after seeing this I think I can replicate it with stacked plywood.

There’s no real reason why a stepped surface wouldn’t work as backloaded designs are often like this. It could be a diffraction nightmare though but I think it’s worth trying