Prescription= BUILD SOMETHING! I’ve been renovating our house for the last few months with no DIY time. Lately, I’ve been buying random audio related parts on EBay which is a sure sign of a need for a build.
So I put together a quick KT88/6550 SE amp. I’ve always wanted to hear what a SE Tube amp sounded like. A piece of plywood for a prototype build, 6n1p driver, external filament and B+ supplies to power it, and a pair of cheap EBay 3.5k transformers.
I built up the first channel and got about what I would expect. About 6 watts max in triode mode cathode biased. Running about 70ma and 400v or so on the plate. At that point I just had to hear it so I set about building a second channel, adding rca inputs and a terminal strip for speaker connections. When I slowly brought up the B+ on the second channel everything initially seemed ok.
I am running 560 ohm RK and the Kathode voltage began to fall as the B+ reached about 200v. The other channel was normal. I stopped at 200v and measured the voltage drop at the output transformer. This indicated that about 2 times the proper current was flowing through the transformer even though the voltage across the RK was lower than normal. I powered down, switched tubes and had the same result.
No DC was appearing on the secondary load resistor so the transformer did not seem to be the issue. The RK measured to spec at 560 ohms. I checked all wiring and then with the tube removed I used an external supply and a series resistor of 10k to check the leakage on the Kathode bypass cap. As I increased the PS voltage I could see the RK voltage increasing and then at about 20 volts it began to go down.
It finally stabilized and using the power supply current and the voltage drop on RK/CK I calculated the impedance at 220ohms. In my 40 years of working in electronics I can’t remember a cap failure like this. I’ve seen lots of them blow up and lots just loose their capacitance but I’ve never seen one do this. It was a used 680uf 250v cap that I probably pulled from a switching PS. I had 2 of them so I used them for L and R Kathode bypasses. They both checked out with identical capacitance. Anyway, I thought it was an interesting failure mode and if that failure had happened in an operating amp it would have caused a lot of problems, so good thing it happened on the test bench. I replaced both of them with new 100v parts.
I set up the amp in my listening room connected to a pair of 95db Pi 3 speakers and driven by an Aikido preamp. Turns out 5 or 6 watts will play pretty loud with these speakers. Bass was a bit different due to the high output Z of about 4 ohms. Overall, the sound was quite pleasant and it easily passed my Dave Brubeck test. If you play Take 5, some part of you should be moving to the music. If not, there’s a problem!
I noticed a bit of softening compared to what I’ve been using, a 100w version of the First Watt F5 turbo, but otherwise it sounded pretty good considering the quick build and cheap transformers. I did not notice confusion or lack of clarity in complex passages that many people claim as a SE tube drawback. I think you need to have efficient speakers with proper impedance and also have realistic expectations for volume. Anyway, lots of experiments, tweaks etc to be done before I build a final version.
Prognosis... Good sound ahead!
Here’s a few pictures. The bench supply has voltage and current meters. I have a 5h choke and 47uf630v solemn cap external to the H.V. supply. This killed the hum completely and I now only hear the preamps small hum when listening right up to the driver. It’s also very quiet as far as hiss. I have been monitoring output voltage levels with a DVM with a fast +and- peak hold for AC signals. 1-2 watts goes a long way for my general listening requirements
PS: The sharp eyed among you will probably notice I am running Svet 6550s. I also have a bunch of EH and Sovtek KT88s- I like the coke bottle shape of the Sovtek. I have a pair of 70s GEC KT88s that I’ve never heard so now I may get a chance to try them.
So I put together a quick KT88/6550 SE amp. I’ve always wanted to hear what a SE Tube amp sounded like. A piece of plywood for a prototype build, 6n1p driver, external filament and B+ supplies to power it, and a pair of cheap EBay 3.5k transformers.
I built up the first channel and got about what I would expect. About 6 watts max in triode mode cathode biased. Running about 70ma and 400v or so on the plate. At that point I just had to hear it so I set about building a second channel, adding rca inputs and a terminal strip for speaker connections. When I slowly brought up the B+ on the second channel everything initially seemed ok.
I am running 560 ohm RK and the Kathode voltage began to fall as the B+ reached about 200v. The other channel was normal. I stopped at 200v and measured the voltage drop at the output transformer. This indicated that about 2 times the proper current was flowing through the transformer even though the voltage across the RK was lower than normal. I powered down, switched tubes and had the same result.
No DC was appearing on the secondary load resistor so the transformer did not seem to be the issue. The RK measured to spec at 560 ohms. I checked all wiring and then with the tube removed I used an external supply and a series resistor of 10k to check the leakage on the Kathode bypass cap. As I increased the PS voltage I could see the RK voltage increasing and then at about 20 volts it began to go down.
It finally stabilized and using the power supply current and the voltage drop on RK/CK I calculated the impedance at 220ohms. In my 40 years of working in electronics I can’t remember a cap failure like this. I’ve seen lots of them blow up and lots just loose their capacitance but I’ve never seen one do this. It was a used 680uf 250v cap that I probably pulled from a switching PS. I had 2 of them so I used them for L and R Kathode bypasses. They both checked out with identical capacitance. Anyway, I thought it was an interesting failure mode and if that failure had happened in an operating amp it would have caused a lot of problems, so good thing it happened on the test bench. I replaced both of them with new 100v parts.
I set up the amp in my listening room connected to a pair of 95db Pi 3 speakers and driven by an Aikido preamp. Turns out 5 or 6 watts will play pretty loud with these speakers. Bass was a bit different due to the high output Z of about 4 ohms. Overall, the sound was quite pleasant and it easily passed my Dave Brubeck test. If you play Take 5, some part of you should be moving to the music. If not, there’s a problem!
I noticed a bit of softening compared to what I’ve been using, a 100w version of the First Watt F5 turbo, but otherwise it sounded pretty good considering the quick build and cheap transformers. I did not notice confusion or lack of clarity in complex passages that many people claim as a SE tube drawback. I think you need to have efficient speakers with proper impedance and also have realistic expectations for volume. Anyway, lots of experiments, tweaks etc to be done before I build a final version.
Prognosis... Good sound ahead!
Here’s a few pictures. The bench supply has voltage and current meters. I have a 5h choke and 47uf630v solemn cap external to the H.V. supply. This killed the hum completely and I now only hear the preamps small hum when listening right up to the driver. It’s also very quiet as far as hiss. I have been monitoring output voltage levels with a DVM with a fast +and- peak hold for AC signals. 1-2 watts goes a long way for my general listening requirements
PS: The sharp eyed among you will probably notice I am running Svet 6550s. I also have a bunch of EH and Sovtek KT88s- I like the coke bottle shape of the Sovtek. I have a pair of 70s GEC KT88s that I’ve never heard so now I may get a chance to try them.
Last edited:
Bfpca,
Great job on the amp!
That is a great system.
And thanks for the excellent writeup.
You stated the output impedance is 4 Ohms. Is that on the 8 Ohm tap?
What were the specifications of your output transformers:
Maximum DC primary current?
Primary inductance?
Primary DCR?
Secondary DCR, on what tap?
Or, did you measure that?
Based on those figures for the output transformer, the most dramatic single change to the amplifier might be if you use different output transformers.
Great job on the amp!
That is a great system.
And thanks for the excellent writeup.
You stated the output impedance is 4 Ohms. Is that on the 8 Ohm tap?
What were the specifications of your output transformers:
Maximum DC primary current?
Primary inductance?
Primary DCR?
Secondary DCR, on what tap?
Or, did you measure that?
Based on those figures for the output transformer, the most dramatic single change to the amplifier might be if you use different output transformers.
Thank You for the complements. My system is completely DIY with a DDDac 1794 as a front end. The OPT were like 30$ each on EBay. I had a coupon as well so they were quite affordable. They are rated at 100ma 8 w 3500-4/8 ohms and they also have a UL tap which I didn’t realize when I ordered them. The response was rated 20-40khz -3db. My own measurements are with it driven by 6550 in triode so the RP is low. Maybe -1db at 20hz and -3db at 26khz. In the low bass distortion sets in early at only 1 watt at 20hz with 60ma dc so I doubt the 100ma rating. The primary DCR is 400ohms and the secondary is somewhere around 1.8 ohms. The voltage ratio worked out to 23.5:1 which looks more like 4400 ohms primary impedance. I did a quick load on/off measurement and got about 4 ohms impedance on the 8 ohm tap.
If you look at RP of 800 ohms and Rt of 400 on primary transformed to the secondary would be 1200/4400x8ohms= 2.2 ohms +1.8 Rsec =4 ohms for 8 ohm tap. The inductance spec was 25h but I have not tried to measure it myself. This is the first step in building a much better SE amp. I threw it together in 2 days. I may try current source loading the 6N1p because the drive voltage needed is 80v p/p. I would rather use ECC99 for a job like this but it just doesn’t have enough gain and I don’t want 3 stages. I could also go with 12ax7 etc with source followers to get more gain. I also have a bunch of 6au7 so a CCS loaded pentode driver and Schade feedback is another alternative to look at. Lots of fun. I have several power transformers around to power it and lots of other bits and pieces. I want to make it an integrated amp with a preamp output for subwoofer use and a switchable high pass filter as I find those 2 features are very useful.
If you look at RP of 800 ohms and Rt of 400 on primary transformed to the secondary would be 1200/4400x8ohms= 2.2 ohms +1.8 Rsec =4 ohms for 8 ohm tap. The inductance spec was 25h but I have not tried to measure it myself. This is the first step in building a much better SE amp. I threw it together in 2 days. I may try current source loading the 6N1p because the drive voltage needed is 80v p/p. I would rather use ECC99 for a job like this but it just doesn’t have enough gain and I don’t want 3 stages. I could also go with 12ax7 etc with source followers to get more gain. I also have a bunch of 6au7 so a CCS loaded pentode driver and Schade feedback is another alternative to look at. Lots of fun. I have several power transformers around to power it and lots of other bits and pieces. I want to make it an integrated amp with a preamp output for subwoofer use and a switchable high pass filter as I find those 2 features are very useful.
Bfpca,
Two stage SE amps are challenging.
I like your idea of using a current source in the driver plate. I use an IXYS current source, the one with a 900V breakdown rating: I do not have to worry about higher B+ as the tubes warm up. At startup, with RC coupling, the current source drives the coupling cap (with 0 Volts across it until it charges up), and the cap drives the grid resistor.
I have used the IXYS with several tube types (the current value is settable).
You could use the IXYS with the 6N1P, u = 33 (I have done that).
The 12AX7 with u = 100 does not have enough current for me to consider it as a driver.
The 12AT7 has u = 60, but is less linear than I prefer.
The 12AY7 has u = 40, and works well with the IXYS current source. I used the two triodes in parallel. I set the IXYS current source to about 6.5 or 7 mA. I had individual self bias resistors with associated bypass caps. Then I used individual grid stopper resistors, and tied the plates together.
That gave reasonable linearity, large voltage swing, and enough current drive for the output stage grid resistor (and capacitance whenever I triode wired the output stage; and when I used DHTs).
Yes, you will probably have to use an unbypassed 12AT7 or a pentode, in order to use the Schade feedback, and still have enough total gain of the amp.
Two stage SE amps are challenging.
I like your idea of using a current source in the driver plate. I use an IXYS current source, the one with a 900V breakdown rating: I do not have to worry about higher B+ as the tubes warm up. At startup, with RC coupling, the current source drives the coupling cap (with 0 Volts across it until it charges up), and the cap drives the grid resistor.
I have used the IXYS with several tube types (the current value is settable).
You could use the IXYS with the 6N1P, u = 33 (I have done that).
The 12AX7 with u = 100 does not have enough current for me to consider it as a driver.
The 12AT7 has u = 60, but is less linear than I prefer.
The 12AY7 has u = 40, and works well with the IXYS current source. I used the two triodes in parallel. I set the IXYS current source to about 6.5 or 7 mA. I had individual self bias resistors with associated bypass caps. Then I used individual grid stopper resistors, and tied the plates together.
That gave reasonable linearity, large voltage swing, and enough current drive for the output stage grid resistor (and capacitance whenever I triode wired the output stage; and when I used DHTs).
Yes, you will probably have to use an unbypassed 12AT7 or a pentode, in order to use the Schade feedback, and still have enough total gain of the amp.
Last edited:
I have used the 450v Ixys before and they work well. I didn’t realize they had a 1000v part. I am going to revamp my PPP kt88 amp at some point and try them in the ECC99 LTP driver and perhaps the 5687 input stage. I built it in 2000 and don’t use it much. I initially ran it triode with no feedback and got 50w with Zout of 2.7 ohms. I then added feedback and had a Z out of .5 ohms but it never sounded as good and had to be used with a preamp due to low gain. I did not have decent equipment when I built it so I am interested in testing it to see how it performs. I had a borrowed scope at the time and using the Hammond 1900 ohm 120w transformers it had flat response to 20khz and could drive 4 ohm /4uf without more than a slight overshoot. I think that particular Hammond offers excellent performance.
I live a few miles from the main road so my power quality is not the best with poor regulation. I regulated all supplies in the PPP amp which was a challenge. I may do the same for the SE amp. It weighs about 100lbs and there is 20lbs of 3/8 and 1/2” 6061 in the chassis. Here it is gathering dust in my workshop. It will be interesting to review what I did nearly 20 years ago and see what I’ve learned since.
I live a few miles from the main road so my power quality is not the best with poor regulation. I regulated all supplies in the PPP amp which was a challenge. I may do the same for the SE amp. It weighs about 100lbs and there is 20lbs of 3/8 and 1/2” 6061 in the chassis. Here it is gathering dust in my workshop. It will be interesting to review what I did nearly 20 years ago and see what I’ve learned since.
Yes, that is certainly an option for getting good drive current capability. There isn't a lot of data out there on driving KT88/6550 into +g1 current. I am surprised at how good this SE amp sounds given the high THD, although it is 2Hd dominant up until near the clipping point. That's the whole point of my experiment. To explore SE sound quality and try a bunch of circuit options. Having a set of speakers that will work with low power is a prerequisite IMO.
I have built 4 or 5 6550 type SE amps using the RH "Universal" circuit with 6sl7 or 6sn7 driver tube or some related tubes. The RH amp operates the 6550 in Pentode mode which is twice as powerful as triode mode. I find these type of amps with plate to plate fb from power to driver tube to be really good sounding amps. When coupled to a high efficiency speaker, i am using Altec 19, the sound quality can be quite enthralling. Several amps with ultralinear taps were examined but i felt that the pure pentode mode to be the best overall. cheers, 808.
Thank you for the advice. Altec 19 - Very nice! I have Pi 3 with upgraded drivers and crossovers in a really well braced box. The lack of strain when using a HE compression driver is something everyone needs to hear. Cymbals etc are so cleanly reproduced. I have seen the RH amps and want to try different output stage configurations. I am now just getting a baseline for comparison.
Eli Duttman,
A 6550 in pentode mode does not require a lot of swing or slew current from a 12AX7. But if Schade feedback is used the 12AX7 will likely run out of current. And the 12AX7 might be more challenged if the 6550 is UL or Triode wired. The effective control grid capacitive reactance is much lower in those cases. Then the MOSFET you mentioned should take care of those more challenging load conditions.
I should have said I would not use a 12AX7 to drive a 300B. Different output tubes, different results.
Yes, the 12AX7 can drive a MOSFET follower driving the 300B. But the 12AX7 might not have enough linear swing if the 300B is biased at 75, 80, or even 90V. And using the same 12AX7 plus the same MOSFET as a common source amp, could provide enough swing, but it would be bandwidth limited because of the current from the 12AX7 would not be able to slew the common source MOSFET gate capacitances.
A 6550 in pentode mode does not require a lot of swing or slew current from a 12AX7. But if Schade feedback is used the 12AX7 will likely run out of current. And the 12AX7 might be more challenged if the 6550 is UL or Triode wired. The effective control grid capacitive reactance is much lower in those cases. Then the MOSFET you mentioned should take care of those more challenging load conditions.
I should have said I would not use a 12AX7 to drive a 300B. Different output tubes, different results.
Yes, the 12AX7 can drive a MOSFET follower driving the 300B. But the 12AX7 might not have enough linear swing if the 300B is biased at 75, 80, or even 90V. And using the same 12AX7 plus the same MOSFET as a common source amp, could provide enough swing, but it would be bandwidth limited because of the current from the 12AX7 would not be able to slew the common source MOSFET gate capacitances.
Last edited:
@DAK808
I did use my Triode PPP KT88 as an integrated amp without feedback for several years. At the time I thought I would try it as a power amp with feedback. It didn’t sound as good with feedback but the output Z went from 2.7ohms to .5ohms which was important at the time because I was running speakers that were 4 ohms and required damping from the amplifier.
I have much better speakers and front end now and my priorities have changed. I will try the Schade style feedback in Pentode mode at some point.
@6A3sUMMER
I would try to avoid 12ax7 if possible. With typically 1ma or less flowing its very likely to be slew limited driving a mosfet, even in source follower mode. I like the ECC99 for a nice linear driver tube that can swing a lot of volts and run at 10+ma. That’s what I used in the LTP in my pp amp driving 2 pairs of KT88s. Without feedback that amp was pretty much flat to 20khz in tride mode with about -45v bias. So 90vp/p drive voltage.
I did use my Triode PPP KT88 as an integrated amp without feedback for several years. At the time I thought I would try it as a power amp with feedback. It didn’t sound as good with feedback but the output Z went from 2.7ohms to .5ohms which was important at the time because I was running speakers that were 4 ohms and required damping from the amplifier.
I have much better speakers and front end now and my priorities have changed. I will try the Schade style feedback in Pentode mode at some point.
@6A3sUMMER
I would try to avoid 12ax7 if possible. With typically 1ma or less flowing its very likely to be slew limited driving a mosfet, even in source follower mode. I like the ECC99 for a nice linear driver tube that can swing a lot of volts and run at 10+ma. That’s what I used in the LTP in my pp amp driving 2 pairs of KT88s. Without feedback that amp was pretty much flat to 20khz in tride mode with about -45v bias. So 90vp/p drive voltage.
FET gyrators are great way to improve amp. definitely worth it. it also increases gain significantly, for high mu-ra toobz. IMHO the jfet is a notch better soundwise.
to Ale Moglia findings.
Yes, a gyrator is another possible. Lots of experiments are possible. Right now the distortion is higher than I would like so first job will be optimizing the driver stage and I’m also going to try fixed bias. I am too busy right now to put much time in on this. I am trying to find some listening time each day to get familiar with the sound before I make changes.
Bfpca,
I am sorry, I do too much "tongue in cheek" writing, and expect everyone to understand the the whole technical discussion, but that is filled with sarcasm. I need to change my style of writing (thanks for calling me out, I will try and remember to change).
If you use the right MOSFET, and the 300B grid bias is set to -35V, the 12AX7 just may barely be able to drive the capacitance and the required +/- 35V swing. Not very well, but it can do it.
I was trying to tell that to someone who 'suggested that the 12AX7 plus a MOSFET could be used as driver'. I wanted to show it depends on what tube you are driving, and what bias that tube is set to.
Just read completely through that last paragraph I wrote in Post # 11 You may then understand I do not like the idea of a 12AX7 driving a MOSFET.
I think I made that clear in my earlier post # 4. I mentioned the lack of current, so another post-er suggested the MOSFET as the solution. I should also have mentioned the lack of linear volts swing, regardless of required current, no MOSFET follower is going to change the lack of linear swing of a 12AX7.
I am sorry, I do too much "tongue in cheek" writing, and expect everyone to understand the the whole technical discussion, but that is filled with sarcasm. I need to change my style of writing (thanks for calling me out, I will try and remember to change).
If you use the right MOSFET, and the 300B grid bias is set to -35V, the 12AX7 just may barely be able to drive the capacitance and the required +/- 35V swing. Not very well, but it can do it.
I was trying to tell that to someone who 'suggested that the 12AX7 plus a MOSFET could be used as driver'. I wanted to show it depends on what tube you are driving, and what bias that tube is set to.
Just read completely through that last paragraph I wrote in Post # 11 You may then understand I do not like the idea of a 12AX7 driving a MOSFET.
I think I made that clear in my earlier post # 4. I mentioned the lack of current, so another post-er suggested the MOSFET as the solution. I should also have mentioned the lack of linear volts swing, regardless of required current, no MOSFET follower is going to change the lack of linear swing of a 12AX7.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.