OK.
Oh well ... LM334Z/NOPB Texas Instruments | Mouser is a very good and inexpensive CCS suitable for this application. You can 'program' it to about 10 mA by using a 6.8 Ω resistor. I've had particularly good luck with these devices. And they're cheap. Buy a 10 pack for under $10.00!
Just saying,
GoatGuy ✓
PS: would you like me to cobble together the circuit diagram? It takes a little time, but I'd do it this afternoon when I return from “the day's Honey, Do!” list.
Also, I forgot to mention... what is the VOLTAGE DROP across the 10 mA CCS, now?
Well heck I have used those underneath a LTP phase splitter . 6SN7 with 27K anode resistors, running 4.2mA . 8 ohm resistor from ground to REF on LM334 to set current. Those were the design parameters anyway. Ill get a couple more and give it a try!
Ill have to measure the new volt drop, after the dog walk, and lunch, LOL
Also, I forgot to mention... what is the VOLTAGE DROP across the 10 mA CCS, now?
With B+ at 375 there was 230 volts on the plates . With B+ 342, still 230 on the plates. That's the CCS doing its job I guess....
so drop was 145, now 112 volts. So now I have ~ 1.12 watts CCS dissipation.
Ill have to measure the new volt drop, after the dog walk, and lunch, LOL
Also, I forgot to mention... what is the VOLTAGE DROP across the 10 mA CCS, now?
With B+ at 375 there was 230 volts on the plates . With B+ 342, still 230 on the plates. That's the CCS doing its job I guess....
so drop was 145, now 112 volts. So now I have ~ 1.12 watts CCS dissipation.
A quad of 6550-size bottles should do more than that. I had commercial quad-8417 amps made 150W stock on a saggy supply, 180W with more B+ caps. Is your big power stage under-fed or mis-matched?
That amp drove the quad big bottles with a 12AU7-type cathodyne-- the big grids are not a heavy load.
I was testing with an 8 ohm load on the 4 ohm taps, less distortion. Ill have to load it down, see what I can get. Output tranny's only good for ~120 watts though. Tried 6550's, distortion nearly doubled. The parallel 6SN7 drive is just an exercise in over-kill. Just playing, seeing how clean I can make it. The power supply is over-kill too, just because...
You might want to tweak the ratio of the two nominal 15k Ra and Rk for the 2nd triode slightly off the 1:1 ratio. The gain from the cathode is not exactly the same as the gain from the anode, which creates some distortion. You could tweak it while looking at the 2015.
BTW The LM334 only takes 40V max, something to be aware of.
Jan
Last edited:
Thanks for the tips! Yes, I thought about putting in a balance pot. I was wondering why distortion is far from equal measured at cathode vs anode. If I use the LM334 it would be under the cathodes, would only see less than 10 volts. I have a pair of amps that use a LM334 under a LTP, put a lot of hours on them (two winters in my home theater), never an issue. It looks like such a frail little part, but it works great!
I would go with 1000uF, and use a cathode CCS such as LM317 or TL783 if you need higher voltages.
If I'm honest though, I have reverted to using an unbypassed cathode resistor except in the case of injecting gNFB to the VA stage of a power amp, or a headphone amp for obvious reasons.
Sounds to me like it’s under-loaded. One pair of big valves will drive a 5k a-a load at 600V B+. Two is overkill for sure. Sure, the distortion will go down as load Z goes up, but then you really have to watch screen current/voltage. You will need to reduce Vg2 to be safe. Once you hit the below the knee and keep driving it, if the plate doesn’t draw more current the screen will. Can be particularly bad with ultra linear, where you can’t turn down Vg2 to compensate.
He's admitted it is mis-loaded, better distortion. (A little odd.)
The OT is good for HUGE power mid-band. The "120W" is for some THD at some low frequency. Unless this is subwoofer-only, I would tend to tube it for 150W for more midrange headroom, but be aware of farting on the lowest notes. (Few commercial recordings have anything below 50Hz.)
Wouldn’t be “a little odd” if it effectively becomes class A. Which further pushes down the available current/power because the load becomes Ra-a/2 not /4. With 4 120’s i’d Use a more normal impedance and bias it up to get the same effect. Alternatively, keep the higher Z and up the voltage to make power, but that brings a host of other problems.
But it’s working, maybe doing what the OP wanted. But keep an eye on the screens - that arrangement could get them glowing white hot with relatively cold plates if you had sustained clipping.
But it’s working, maybe doing what the OP wanted. But keep an eye on the screens - that arrangement could get them glowing white hot with relatively cold plates if you had sustained clipping.
Output transformer is a Hammond 1650T, primary impedance is 1.9K. Ultralinear taps to screens through 220 ohm resistors. Biased to 60mA, B+ to output tranny about 485V under load. Im not well versed on calculating load lines and output tube operating parameters as you can tell! I wont be using it for low end, I bi-amp and use a 100+lb sand amp for bottom end, LOL. In my system I will use 4 ohm speakers, on the 4 ohm taps. Although after hearing this thru a 12" full range Eminence in an infinite baffle, I' wondering if I did my home theater all wrong ! The bass was surprising good, and the sound was so much more coherent.
PS, I figured the lower distortion with a lighter load was normal tube behavior, IE, they don't like being loaded down. But then the output tubes are pentodes, and the behavior of pentodes is a bit of a mystery to me. They behave one way connected as triodes, another way with regulated screens, an another way in ultralinear.
PS, I figured the lower distortion with a lighter load was normal tube behavior, IE, they don't like being loaded down. But then the output tubes are pentodes, and the behavior of pentodes is a bit of a mystery to me. They behave one way connected as triodes, another way with regulated screens, an another way in ultralinear.
Last edited:
Depends on the pentode. Some really do have less distortion with higher Z loads. Some have a clear optimum. And sometimes you’re interested in maximum power or use of iron, other times it’s secondary.
Coherent sound is a characteristic of good full range drivers. Even if the frequency response isn’t flat. Multi way can be coherent but it NEVER gets that way by accident.
Coherent sound is a characteristic of good full range drivers. Even if the frequency response isn’t flat. Multi way can be coherent but it NEVER gets that way by accident.
Unless you know what harmonics these are, THD numbers are next to useless.
If it's 2nd and 3rd, those will be almost inaudible. But if they're 7th and up, probably audible.
What's the sensitivity of your speakers? Where's your noise floor? What's your target peak SPL? These set how often you'll get into these levels of distortion.
<rant>
Read your Hamm and Crowhurst. JAES 1973 21(4) and 1957 5(4) respectively.
</rant>b
If it's 2nd and 3rd, those will be almost inaudible. But if they're 7th and up, probably audible.
What's the sensitivity of your speakers? Where's your noise floor? What's your target peak SPL? These set how often you'll get into these levels of distortion.
<rant>
Read your Hamm and Crowhurst. JAES 1973 21(4) and 1957 5(4) respectively.
</rant>b
This is the big compromise - pushing it into A1 will drop watts but clean up everything.
Last edited:
I got similar THD readings with a 1650TA OPT but with 18dB of NFB. I used six EL34's and a similar driver arrangement as yourself. I used ECC99's/6H6N's paralleled with a MJE340 CCS from a negative supply of -100v (80v loaded). The OP stage is direct biased adjustable. Grid R's are 47k with grid leak R's of 2k2, g2 R's 220r.
A balance pot is an excellent idea, I used a ganged 10k lin pot/s, makes balance adjusting less fiddly. I used two of those little 4mm banana test sockets on the chassis to make nulling easier. You basically connect your DMM from g1 to g1 and adjust/null trying to get as near 0v as poss. I used 1 turn Bourns pot/s, 10 turn would have made adjustment easier as you can't get it bang on but it's not really needed.
The Hammond 1650TA is a good OPT, I got 145w out of my OP stage albeit briefly, the OPT didn't get hot. Bass response is good down to 5hz, flat to 30khz so a decent not expensive OPT.
Re matching you can play with the value of your dummy load to see if it improves things, try lowering it a tad to say 4ohms on the 8 ohm tap or increase it. The old datsheets for the EL34 and KT88 made matching a lot easier as typical anode resistance was given for various conditions, the Tungsol datasheet http://www.tungsol.com/specs/kt120-tung-sol-specs-curves.pdf - I found didn't have these figures, it's a bloody joke as far as datasheet's go. The Valve Wizard site has some good info on working out anode resistance for beginners, more thorough analysis is in Crowhurst and The Radio Designers Handbook.
I'd advise fitting a protection circuit that cuts power to the mains tfmr if Ik on any valve is exceeded, also fit reverse diodes from anode to ground to protect the OPT if no load is attached, a 1k resistor across the OP terminals does the same job to a point.
Lastly make sure to test for stability, this is the trickiest aspect to get right but with a few tricks not too hard.
Good luck, Andy.
A balance pot is an excellent idea, I used a ganged 10k lin pot/s, makes balance adjusting less fiddly. I used two of those little 4mm banana test sockets on the chassis to make nulling easier. You basically connect your DMM from g1 to g1 and adjust/null trying to get as near 0v as poss. I used 1 turn Bourns pot/s, 10 turn would have made adjustment easier as you can't get it bang on but it's not really needed.
The Hammond 1650TA is a good OPT, I got 145w out of my OP stage albeit briefly, the OPT didn't get hot. Bass response is good down to 5hz, flat to 30khz so a decent not expensive OPT.
Re matching you can play with the value of your dummy load to see if it improves things, try lowering it a tad to say 4ohms on the 8 ohm tap or increase it. The old datsheets for the EL34 and KT88 made matching a lot easier as typical anode resistance was given for various conditions, the Tungsol datasheet http://www.tungsol.com/specs/kt120-tung-sol-specs-curves.pdf - I found didn't have these figures, it's a bloody joke as far as datasheet's go. The Valve Wizard site has some good info on working out anode resistance for beginners, more thorough analysis is in Crowhurst and The Radio Designers Handbook.
I'd advise fitting a protection circuit that cuts power to the mains tfmr if Ik on any valve is exceeded, also fit reverse diodes from anode to ground to protect the OPT if no load is attached, a 1k resistor across the OP terminals does the same job to a point.
Lastly make sure to test for stability, this is the trickiest aspect to get right but with a few tricks not too hard.
Good luck, Andy.
Last edited:
That's pushing it a bit - single-tone tests are very limited, in any realistic signal intermodulation distortion is an issue, and 2nd and 3rd order IM products are as audible as 7th or 8th order (for the same amplitude), they all can fall at non-harmonically related frequencies.
It's my understanding that triodes produce primarily 2nd order harmonics, with pentodes you start introducing 3rd order, higher harmonics are created when you use feedback. Its also my understanding that our ear/brain doesn't mind the even order harmonics but the odd order harmonics sound harsh. I usually just measure 2nd and 3rd order, because those will be the primary harmonics in a tube amp with no feedback. If I get 2nd and 3rd as low as possible without feedback, it stands to reason the higher harmonics will be as low as I can get them also. I can measure way up the spectrum, and measure individual or defined bands of harmonics, but at this point that wouldn't help me much. If I get 2nd and 3rd as low as I can, the rest will follow. The speakers it will eventually power are about 90 db efficient, my test speaker is around 97. Noise is nearly inaudible with the 97 db efficient speaker, so it will be fine with my home theater mains. I don't listen nearly as loud as I used to , I'm not worried about running out of power. I'm just shooting for low distortion. I'd like to get as much Class A power as possible, but at this point I don't know how to go about that, except to bias the outputs as high as safely feasible.
Maybe I should lower my grid resistors, Im at 100K on each tube.
Today I think I'll install a balance pot. It was near perfect with the Sylvania Bad Boys, but with the Russians in there its off a bit. I like the test socket idea, Id hate to have to flip these things over every time I wanted to test that in the future. I would mess around and try and draw a load line for the outputs but with a real life speaker/crossover load the impedance the amp sees is all over the place anyway so....also Im not going to change my output tranny or my B+ at this point. The bias point would be the only thing I can really adjust. I will probably see what effect that has on distortion.
I do have an old Crown IMD tester, practically NOS. But I probably wont go there, LOL
I ordered some 8.2 volt zeners to put under the paralleled 6SN7's at the output of my driver stage, instead of the 830 ohm resistors, per Morgan Jones. Cant wait to pop them in and see what happens with the distortion.
Why not try a 9V battery for grid bias?
Good idea! I used battery bias in a phono preamp some years back, but forgot all about it until now, LOL
Heres basically what the amps will look like when done. In this pic, I had Lundahl output transformers, but it turned out the power loss was quite high in my application.
An externally hosted image should be here but it was not working when we last tested it.
I have to use the "open in a new window" option to see them myself. Not sure why they don't show up in the post.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.