spiggs,
Would it be possible to update the list of modifications/values you have ended up with at this point? (If not I will reread your posts).
I’m ready now to put the rest of my UNSET amp together, and will be grateful to know your final status and operating conditions. I will use 26LW6 and 12GN7a drivers and the single version of 10M45 CCS, with my B+ probably close to 450Vdc. Unfortunately I’m not REW savvy (yet), so I will not be able to measure my results like you did.
Would it be possible to update the list of modifications/values you have ended up with at this point? (If not I will reread your posts).
I’m ready now to put the rest of my UNSET amp together, and will be grateful to know your final status and operating conditions. I will use 26LW6 and 12GN7a drivers and the single version of 10M45 CCS, with my B+ probably close to 450Vdc. Unfortunately I’m not REW savvy (yet), so I will not be able to measure my results like you did.
I have this one Rubycon 500VXG47MEFCSN22X25. Did not see any surge voltage specs listed. Good idea to add an inrush current limiter. I have a soft start board I am using but perhaps a CL-60 before it would tame it a little more.
440Vac*1.414Vpk=622.16Vac
Considering your rectifier will have an additional 30Vdc+ drop, you should be fine.
In-rush current limiters are still a good idea nonetheless.
Considering your rectifier will have an additional 30Vdc+ drop, you should be fine.
In-rush current limiters are still a good idea nonetheless.
@Tubelab_com
extending the concept of the "distortion-only feedback" I proposed on the "shade" thread, would you think that something similar would work?
Here is just a block schematic to explain the idea: will it keep the gain of the pentode, with the rp of the UNSET?
extending the concept of the "distortion-only feedback" I proposed on the "shade" thread, would you think that something similar would work?
Here is just a block schematic to explain the idea: will it keep the gain of the pentode, with the rp of the UNSET?
Here is how my UNSET is currently setup
Antek AS-3T350
Choke – 5H 300mA
C2 – 3 X 100uF 440VAC Motor Run Cap
C1 – 50uF 440VAC Motor Run Cap
OPT KT88SE using 8 ohm tap into 8 ohm or KT88PSE using 4 ohm tap into 8 ohm
Driver tube 12GN7a
Output tube 26HU5
Changes from BOM values or build notes
R7 – 3.4K ohm
R8 – 75K ohm
R113/213 – 15K ohm
D4 - 100V zener
CCS added in place of R108/208 using a IXCP10M45S current regulator tested to give 10mA using a 330 ohm resistor plus 7.5K ohm resistors in parallel to reach desired value. 1K ohm for the gate stopper and a 1M ohm across R108/208 under the CCS.
Feedback mod using a 1M ohm resistor between the junction of R114 and R115 and junction of R105, R106 and R107.
B+ with this setup is about 430V. I am running 250V on the driver tube plate and have the output tubes biased at 103mA. Using the KT88PSE OPTs I get 23 watts of output, with the KT88SE I am seeing 21 watts before clipping. Either OPT will make 20 watts in the 1.5 to 1.8% THD range.
Marvelous! Thank you! Very clear, concise and convenient. My power transformer is an Antek AS-4T325; I hope it will get B+ close to 430Vdc.
Couple of small questions.
1. When you say “I am running 250V on the driver tube plate” is that measured (at min signal) between ground and the plate (pin 7) for the 12GN7a, i.e. after the CCS? What voltage is the cathode sitting at?
2. What is the voltage to ground at C7. (With that difference we could calculate the effective resistance of the CCS at the operating point.)
3. With the new “big” choke installed, do you still experience a sag in B+ level as the UNSET amp nears maximum power? Have you quantified the current sag if any? IIRC you also observed a drop in power tube screen voltage (that is ”fixed” by D3 and Q1) at max power output. What could cause this? Does that mean the B+ dropped enough to bring D3 out of Zener regulation?
Couple of small questions.
1. When you say “I am running 250V on the driver tube plate” is that measured (at min signal) between ground and the plate (pin 7) for the 12GN7a, i.e. after the CCS? What voltage is the cathode sitting at?
2. What is the voltage to ground at C7. (With that difference we could calculate the effective resistance of the CCS at the operating point.)
3. With the new “big” choke installed, do you still experience a sag in B+ level as the UNSET amp nears maximum power? Have you quantified the current sag if any? IIRC you also observed a drop in power tube screen voltage (that is ”fixed” by D3 and Q1) at max power output. What could cause this? Does that mean the B+ dropped enough to bring D3 out of Zener regulation?
Roberto, this will never work! The paper is too wrinkled! 😀
More seriously, I will have to read the other thread - I assume you explained the functioning there. But for the record, what improvements would this achieve in the UNSET if it works?
🙄 I had to admit I take a first photo, then trashed the paper, then realized I forgot one resistor 😒
So I tried to recover the paper and... well... that's the result.
I'm trying that feedback method on a PP amp by now, and I want it to be switchable.
You have to design the amp to sound good without gnfb, and then apply this nfb method.
The benefit is that is the amp has no distortion, you have no feedback applied, so you can stick on two gain stages.
If there's any distortion, the feedback will correct it proportionally (amplified by the gain of the nfb gain stage).
What you get there is a very low THD, an high DF with no extra gain stages needed on the gain path (except the one on the nfb, that works at very low swings, so very clean).
Applying it here, it should be possible to get pentode-like gain, unset-like Rp, low distortion and high DF.
So I tried to recover the paper and... well... that's the result.
I'm trying that feedback method on a PP amp by now, and I want it to be switchable.
You have to design the amp to sound good without gnfb, and then apply this nfb method.
The benefit is that is the amp has no distortion, you have no feedback applied, so you can stick on two gain stages.
If there's any distortion, the feedback will correct it proportionally (amplified by the gain of the nfb gain stage).
What you get there is a very low THD, an high DF with no extra gain stages needed on the gain path (except the one on the nfb, that works at very low swings, so very clean).
Applying it here, it should be possible to get pentode-like gain, unset-like Rp, low distortion and high DF.
1 - I measure the plate voltage at the base of C103/203 referenced to ground with no signal.
2 - I'll have to measure this when I get back home in a few days.
3 - From what I understand the resistor on the output tube screen reg is designed to bring the voltage down as the amp nears clipping to provide both a softer clipping and protection from runaway current spikes. With the smaller choke the amp was fairly sensitive to changes in this resistor. With the big choke in place I find that the 2K value works fine, screen voltage holds steady up to just before clipping and then starts to just drop off by a few volts. Lower values keep the voltage steady but I don't see any more power from the amp. With the bigger choke B+ remains steady up to clipping.
I have no idea if it will work in the real world, but there aren't too many opamps out there that will swing the voltage needed to directly drive the grid of a medium to high powered sweep or other audio output tube. If you decide to breadboard something like this, I would start small. Test things with little tubes first. Some of my early UNSET experiments used the 50C5 radio tube. It is like a smaller version of the 6W6, but still needs about 20 to 25 volts P-P of drive voltage to get 20 watts from a pair in push pull. There is probably a similar 6 volt tube in Europe but I don't know the number offhand. The 6AQ5 is common worldwide, but it needs about 30 volts P-P of grid swing.
My UNSET board has not seen power in several weeks due to preparations for the hamfest, and the cleanup operation involving putting away everything that I did not sell. Nobody offered me any more than $25 each for my 4D32 transmitting tubes, so I decided to keep them and answer the still unanswered question in my head. What will they do in the UNSET circuit? The conservative 50 watt plate dissipation rating hints at big power, but the curves don't look too promising in the efficiency department. Guess there's only one way to find out what will happen.
The driver in the current UNSET board walks a fine line between not quite enough gain with a small plate load resistor, and poor linearity with a large value resistor due to low plate voltage. The CCS eliminates that issue by allowing for a different AC plate load impedance and DC plate load resistance. The effective DC resistance of the CCS is in the 15K ohm range (assuming 150 volts and 10 mA), but it's AC dynamic impedance is in the 500K to 1 meg range at 10 mA allowing for more stage gain.
The average plate current in a pure class A stage should stay relatively constant with an increase in drive level until clipping is approached. The UNSET design with the feedback mod cheat's this a bit by shifting the bias current upwards as the drive is increased. This means that at somewhere around 15 watts with our current 26HU5 - 26LW6s on 400 volt B+ the output the tube current will start to increase. This current increase can result in a sag in B+ due to power transformer or rectifier tube voltage drop. I have a Hammond 272JX (300-0-300 volt) transformer with solid state diodes in my amp currently. The B+ is just under 400 volts at 20 watts output, with both channels driven.
The resistor in series with Q1 is what causes the screen voltage to drop at full power output. It should be sized to keep the screen voltage constant until the amp is driven hard enough so that you wouldn't want to listen to it anyway. This makes life easier on Q1 and may save a tube in the event of a shorted cathode mosfet or unconnected plate cap.
I've sketched an OPAMP because it's what I'm trying with the PP amp, where I need a small swing to achieve good results, but here most probably a triode or nfet will work better (scaled down output to the grid and input signal to the cathode).
Thanks, I will for sure.
Back home so had some time in the garage. C7 measures 344VDC
Wired in a CL-60 into the primary and now the maximum I see on startup is 475VDC. C1 is back to the 500V electrolytic cap.
Thanks for that voltage measurement. I will watch mine when I fire it it, hopefully sometime in next week. I’m doing a choke instead of R7, so I will monitor this voltage closely.
Sounds like the CL-60 nicely fixed the over-voltage at startup.
Sounds like the CL-60 nicely fixed the over-voltage at startup.
Yes, it tamed the voltage spike nicely. A little disappointed though that the soft start board I installed didn't do the job by itself. It has a 2 second delay from turn on switching between a current limiter for the soft start to direct pass through. Doesn't seem enough in this case. I put the CL-60 before the soft start board but now am thinking perhaps it would be better after so it only limits the voltage going into the transformer and not the smps that I am using for the heaters.
I realized I made an error here. I was using 75K ohm for R8 with the 56V zener in place and 42K ohm with the 100V zener. Remembered this as I was updating the UNSET board I have configured for 6EJ7/6DQ5 tubes. With the 6EJ7 I am going to try a CCS setup for 5.7mA in combination with the feedback mod and other changes. I am going to try for a more compact build with this one using a 290uF electrolytic for C2, a smaller Hammond 159T choke, and the heaters running off the 6V windings on the toroidal. If it's quiet enough for headphone and near field speaker use I could set it up as my office amp. If not I'll probably use it as my garage amp.
Hi Spiggs, the Antek AS-3T350 has a 6.3v@4A secondary. Doesn't the 26HU5 need a 26v heater?
I have a pair of 26LW6 and I'm confused with the heater voltage and current requirements 🙁
I have a pair of 26LW6 and I'm confused with the heater voltage and current requirements 🙁
That is correct. For my 26HU5 tubes I am using a separate Meanwell RS-75-24 SMPS for the 26V heater supply. I have also used a 25VAC transformer and that works just as well. Spec sheet says 0.6A per tube so 1.2A for the pair. With the SMPS I need to add an inrush current limiter to the output so the SMPS doesn't go into hiccup mode at startup even with its 3.2A rating. The 25VAC transformer is rated at 2A and of course has no problem starting up the heaters.
Note with the AS-3T350 you can configure the two 6.3V secondaries to give 12V and that works well with the 12V heater driver tube choices if a bit overkill since they only need 300mA each.
Thanks Spiggs. I can get custom transformers made here, been using them for all Tubelab designs.
Would these specs be fine for a B+ around 420-440v ?
350-0-350v@250ma
0-26v@2A
Would these specs be fine for a B+ around 420-440v ?
350-0-350v@250ma
0-26v@2A
I would go for something with some headroom on the 350-0-350 winding. 300 or 350mA The 26V at 2A should be good for the output tube heaters. What driver tubes are you using? You'll need a 6.3 or 12.6V 1A winding for those depending. Are you using a tube or solid state rectifier? If a tube you will also need to provide heater voltage for that as well.