HPA - Glow One (diy KT88 amplifier)

[horias2000]

After building several solid state amplifiers, I stared reading about vacuum tube amplifiers. I already knew that the technology is really outdated and no matter how much we try with tubes, transistors will always measure and sound better. I'm an electronics engineer so I do not really buy into the "tube sound" thing. But nevertheless, I was curios to learn more about these devices from the past and I started reading some books and articles (Morgan Jones and others). As I was reading I stared having an itch to actually build one. As I was reading more and more, the itch became unbearable. I then decided that I must design and build one tube amp. The entire process I went through can be seen here (https://www.audiosciencereview.com/...g-a-valve-amplifier.29903/page-4#post-1385454) . I will not go through the entire process, I just want to make a separate thread with the final measurements of the amplifier. The staring point of the project was the design by Bob Cordell (C70) that was published on Linear Audio. I decided to use new tubes (JJ) and new toroidal transformers (Toroidy from Poland). The final amplifier has a maximum output power of about 2x 20W. I also added a milliamp meter so that I can measure the current going through the KT88s.

The final stage of the amplifier is in pentode mode. I designed the amp so that I can test UL mode as well.
I have a QA401 that I used for measurements. The setup was pretty straight forward. QA401 output -> Power amp -> QA401 input. I used a resistive load of 8OHMs made out of 50W wirewound power resistors.

I will start with the usual measurements:

1W into 8OHMs:

5W into 8OHMs:

Multitone (QA401 software is not that good for multitone. It can't increase/decrease the level of the signal):

Frequency response for the 8OHM tap:

THD at 1W and 5kHz

THD at 1W and 10kHz

THD at 1W and 20kHz

THD at 1W at 50Hz

Some IMD measurements:

Now 1W into 8OHMs in UL mode:

I have to say that the measurements were a lot better than what I was expecting. Having a THD+N of 77dB at 1W is really good in my opinion. Most of my listening happens at 1W or below. I know that there are solid state amplifier that measure a lot better but I must say that I was pleasantly surprised by the results. The only "bad" thing is that UL mode measures a lot worse than pentode. This means that the finished amp will stay in pentode mode. I also have to say that I worked a lot on the amp to make it stable (thanks to @SIY and other forum members ) and I saw only measurements until it was almost done. I only stared listening to the thing a month ago or so. From the listening point of view I have to say that it sounds pretty well (as expected from the measurements). I did not have the chance to listen to it in UL mode as it was already setup for pentode mode as UL had significantly worse results. Attached a few pictures with the amp itself. I think it came out really ok . Next I will post similar measurements with my solid state integrated amplifier but only after I solve some pesky ground loops inside the amp .

 

[lashto]

very good measurements for a tube amp.
I'd like to see a lot more measurements of tube gear but unfortunately they are not particularly popular on ASR

 

[horias2000]

very good measurements for a tube amp.
I'd like to see a lot more measurements of tube gear but unfortunately they are not particularly popular on ASR

Yes, the amp measures really well considering other tube amplifiers I've seen. Commercially, I could only find a few that had a THD of 0.03% at 1W into 8OHM. I think this is mostly due to the fact that they intentionally want more THD in order to sound more "tube like". I do not agree with this philosophy as I do not want my amp to not add anything to the original recording. In reality any amp will have some degree of non-linearity, but it should be kept as low as possible. I personally like how this tube amp sounds but I have to agree that solid state amps are considerably more accurate and measure considerably better. THD might be ok for this tube amp but if you look at IMD, the tube amp is far behind a good solid state amp. Fortunately I now have both the tube amp and the solid state one and I can listen to the one I want

 

[lashto]

Yes, the amp measures really well considering other tube amplifiers I've seen. Commercially, I could only find a few that had a THD of 0.03% at 1W into 8OHM. I think this is mostly due to the fact that they intentionally want more THD in order to sound more "tube like". I do not agree with this philosophy as I do not want my amp to not add anything to the original recording. In reality any amp will have some degree of non-linearity, but it should be kept as low as possible. I personally like how this tube amp sounds but I have to agree that solid state amps are considerably more accurate and measure considerably better. THD might be ok for this tube amp but if you look at IMD, the tube amp is far behind a good solid state amp. Fortunately I now have both the tube amp and the solid state one and I can listen to the one I want

I like tube amps in general and the tube THD sounds good to me.

Judging by the HD-spectra your amp does not use NFB (or uses very little of it). Tube amps using lots of NFB to lower distortion might be missing the point, IMO. You might reach 80-90dB SINAD but you won't convince any SINAD fans with that number. And a tube amp that sounds and measures like an SS loses in every other aspect (power/price/maintenance/etc) so why would anyone buy it?
Just leave the tube THD as is and the fans will buy it for its "tube sound" and be happy. And SINAD fans will buy their 120dB amps and be happy too
Everybody happy

 

[horias2000]

I like tube amps in general and the tube THD sounds good to me.

Judging by the HD-spectra your amp does not use NFB (or uses very little of it). Tube amps using lots of NFB to lower distortion might be missing the point, IMO. You might reach 80-90dB SINAD but you won't convince any SINAD fans with that number. And a tube amp that sounds and measures like an SS loses in every other aspect (power/price/maintenance/etc) so why would anyone buy it?
Just leave the tube THD as is and the fans will buy it for its "tube sound" and be happy. And SINAD fans will buy their 110-120dB SS amps and be happy too
Everybody happy

The amp has moderate NFB. About 20dB of NFB. I think that NFB is mandatory if you want to have a decent amplifier. I see no valid reason to not have NFB. I don't necessarily agree with the HD approach you mention. As I said, I do not want my amp to add anything to a recording. I want to hear it as the artist/mastering engineer intended it. Having lots of HD is just poor design in my opinion. If Eric Clapton decided to add a lot of distortion to his guitar than that is what I want to hear and not any extra HD added by the amp. This amp can also work in UL mode but in this mode the THD was around 0.2%. This for me is unactable if I know that I can get 0.01% in pentode mode. I know that back in the day, tube amps with high distortion were normal but nowadays we should not accept bad performance. If I want extra THD I can add that from my DAC

 

[lashto]

The amp has moderate NFB. About 20dB of NFB. I think that NFB is mandatory if you want to have a decent amplifier. I see no valid reason to not have NFB. I don't necessarily agree with the HD approach you mention. As I said, I do not want my amp to add anything to a recording. I want to hear it as the artist/mastering engineer intended it. Having lots of HD is just poor design in my opinion. If Eric Clapton decided to add a lot of distortion to his guitar than that is what I want to hear and not any extra HD added by the amp. This amp can also work in UL mode but in this mode the THD was around 0.2%. This for me is unactable if I know that I can get 0.01% in pentode mode. I know that back in the day, tube amps with high distortion were normal but nowadays we should not accept bad performance. If I want extra THD I can add that from my DAC

sounds like you are a SINAD fan building tube amps . I am a bit surprised that the amp still has a very tube-like HD spectrum (i.e. f-noise & H2 dominated) at 20dB NFB. Kudos for the impl!

And talking about the distortion, it's not a 100% given that you'll hear less HD from a 120dB SINAD amp than from a 50dB SINAD amp. What reaches your ears is the combined HD of amp & speakers and those H-s can also cancel each other. Here's an interesting paper that details the effect. Very slim chances but still...

Generally, your speakers add a lot more THD than all electronics combined. In terms of total system-SINAD, the difference between the 'cleanest' SS amp and the 'worst' tube amp is below 1dB SINAD (probably closer to 0.1). And that's the happy case for the SS amp. The happy case for a tube-amp (i.e. HD cancellation) can actually bring you much closer to what Eric Clapton (supposedly) wanted.

So, zero chances with SS vs. slim chances with tubes .. tough choice

 

[horias2000]

sounds like you are a SINAD fan building tube amps . I am a bit surprised that the amp still has a very tube-like HD spectrum (i.e. f-noise & H2 dominated) at 20dB NFB. Kudos for the impl!

And talking about the distortion, it's not a 100% given that you'll hear less HD from a 120dB SINAD amp than from a 50dB tube one. What reaches your ears is the combined HD of amp & speakers and those H-s can also cancel each other. Here's an interesting paper that details the effect. Very slim chances but still...

Generally, your speakers add more THD than all electronics combined. In terms of total system-SINAD, the difference between the 'cleanest' SS amp and the 'worst' tube amp is below 1 SINAD (probably around 0.1). And that's the happy case for the SS amp. The happy case for a tube-amp (i.e. HD cancellation) can actually bring you much closer to what Eric Clapton (supposedly) wanted.

So, zero chances with SS vs. slim chances with tubes .. tough choice

I don't consider myself a SINAD fan but being an engineer, I can't help but try to achieve the best performance I can (this makes me a bit of a SINAD fan ). I think that having a sinad >70dB is enough to no be bothered by it. I agree that speakers contribute a lot more to HD than the electronics. Good speakers can have an THD of around 0.1 - 0.2% which is a lot more than a good amp. I know that HD components can cancel themselves out but at the same time these can also add on top of each other. So, I think that keeping the THD of the electronics to a decently low value is a must.

In the measurements of the tube amp I made the power supply components are a bit larger than reality. The 50Hz component is actually around -90dB. I realized that the coax cable going into the analyzer was close to the soldering station and it was picking up a lot of stray 50Hz stuff. I was too lazy to redo all the measurements

 

[lashto]

In the measurements of the tube amp I made the power supply components are a bit larger than reality. The 50Hz component is actually around -90dB. I realized that the coax cable going into the analyzer was close to the soldering station and it was picking up a lot of stray 50Hz stuff. I was too lazy to redo all the measurements

If you want bragging rights, just do another round. Who knows, maybe you can "win" this thread

 

[SIY]

This amp can also work in UL mode but in this mode the THD was around 0.2%. This for me is unactable if I know that I can get 0.01% in pentode mode.

That seems like a large spread. Perhaps the UL tapping or anode-anode load impedance isn't optimal for UL with that tube?

 

[Gorgonzola]

After building several solid state amplifiers, I stared reading about vacuum tube amplifiers. I already knew that the technology is really outdated and no matter how much we try with tubes, transistors will always measure and sound better. I'm an electronics engineer so I do not really buy into the "tube sound" thing. But nevertheless, I was curios to learn more about these devices from the past and I started reading some books and articles (Morgan Jones and others). As I was reading I stared having an itch to actually build one. As I was reading more and more, the itch became unbearable. I then decided that I must design and build one tube amp. The entire process I went through can be seen here (https://www.audiosciencereview.com/...g-a-valve-amplifier.29903/page-4#post-1385454) . I will not go through the entire process, I just want to make a separate thread with the final measurements of the amplifier. The staring point of the project was the design by Bob Cordell (C70) that was published on Linear Audio. I decided to use new tubes (JJ) and new toroidal transformers (Toroidy from Poland). The final amplifier has a maximum output power of about 2x 20W. I also added a milliamp meter so that I can measure the current going through the KT88s.

The final stage of the amplifier is in pentode mode. I designed the amp so that I can test UL mode as well.
I have a QA401 that I used for measurements. The setup was pretty straight forward. QA401 output -> Power amp -> QA401 input. I used a resistive load of 8OHMs made out of 50W wirewound power resistors.

I will start with the usual measurements:

1W into 8OHMs:
View attachment 244745

5W into 8OHMs:
View attachment 244746

Multitone (QA401 software is not that good for multitone. It can't increase/decrease the level of the signal):
View attachment 244748

Frequency response for the 8OHM tap:
View attachment 244751

THD at 1W and 5kHz
View attachment 244761

THD at 1W and 10kHz
View attachment 244762

THD at 1W and 20kHz
View attachment 244763

THD at 1W at 50Hz
View attachment 244764

Some IMD measurements:
View attachment 244765

View attachment 244766

View attachment 244767

Now 1W into 8OHMs in UL mode:
View attachment 244771

I have to say that the measurements were a lot better than what I was expecting. Having a THD+N of 77dB at 1W is really good in my opinion. Most of my listening happens at 1W or below. I know that there are solid state amplifier that measure a lot better but I must say that I was pleasantly surprised by the results. The only "bad" thing is that UL mode measures a lot worse than pentode. This means that the finished amp will stay in pentode mode. I also have to say that I worked a lot on the amp to make it stable (thanks to @SIY and other forum members ) and I saw only measurements until it was almost done. I only stared listening to the thing a month ago or so. From the listening point of view I have to say that it sounds pretty well (as expected from the measurements). I did not have the chance to listen to it in UL mode as it was already setup for pentode mode as UL had significantly worse results. Attached a few pictures with the amp itself. I think it came out really ok . Next I will post similar measurements with my solid state integrated amplifier but only after I solve some pesky ground loops inside the amp .

Really nice, elegant construction. Performance (to my inexpert eye) looks really good, in the league of the Sonic Frontiers SFS-80.

 

[GXAlan]

Meaaures great! The QA401 measures worse than an APx555 /E1DA Cosmos, so you can add a few dB to your scores.

The JBL SA600 got 66 dB 5W SINAD on a QuantAsylum and Amir got to 71dB.

 

[valve_integrator]

"After building several solid state amplifiers, I stared reading about vacuum tube amplifiers. I already knew that the technology is really outdated and no matter how much we try with tubes, transistors will always measure and sound better."

I am sorry you are wrong.
I must also suspect you did all your measurements into a dummy load (a resistor).
Very different things happen once connected to real loudspeakers (especially ESLs!)

We have done many double blind tests, including evaluating whether or not a "hi end" shop owner could tell the difference between hi rate mp3, low rate mp3 and uncompressed 24 bit wav.
(what was so suprising from the so called "golden ears boys" is they preferred the mp3 version, with all those masking artefacts, and could scarcely tell whether or not a DSP was in use to filter off some nasty room resonances)..

I therefore very strongly question whether or not you have the ability to hear the difference between the technologies you claim to, and in reality the above statement was made out of a mix of ignorance or contempt for a subject you knew little about.
Whilst I do accept (having tested loads of them!) that the majority of mass produced valve and sand based amps leave a lot to be desired I simply have to rely on the results of a week of (mostly blind) tests, we did stacking a "so called" hi end solid state system with all the owner's favourite tweaks, giving us a dead spot-on comparison between JBL and another well known marque's speakers...and then driving them with a mix of music from his system, my own design of amp (which took about 5yrs to decide on the design!) and an ancient but well restored pair of USA made monoblocks from 1947.
There was no doubt in anyone's mind at the end of our marathon listening sessions, that the design I had come up with sounded qualitively much better than the others and made all this digital material sound "good" even when the recording and bit rates were far from optimal.
(FYI we also used some old AKM convertors in the studio grade DAC, compared with commercial DACs with some "tuby plugins/options, and even with a "tube sound" preamp internally equipped with twin triodes...)

Valve amps "can" make this harsh digital sound a quantum leap better and less tiring on the ears, but you don't start by making statements from pure prejudice.

 

[SIY]

"After building several solid state amplifiers, I stared reading about vacuum tube amplifiers. I already knew that the technology is really outdated and no matter how much we try with tubes, transistors will always measure and sound better."

I am sorry you are wrong.
I must also suspect you did all your measurements into a dummy load (a resistor).
Very different things happen once connected to real loudspeakers (especially ESLs!)

We have done many double blind tests, including evaluating whether or not a "hi end" shop owner could tell the difference between hi rate mp3, low rate mp3 and uncompressed 24 bit wav.
(what was so suprising from the so called "golden ears boys" is they preferred the mp3 version, with all those masking artefacts, and could scarcely tell whether or not a DSP was in use to filter off some nasty room resonances)..

I therefore very strongly question whether or not you have the ability to hear the difference between the technologies you claim to, and in reality the above statement was made out of a mix of ignorance or contempt for a subject you knew little about.
Whilst I do accept (having tested loads of them!) that the majority of mass produced valve and sand based amps leave a lot to be desired I simply have to rely on the results of a week of (mostly blind) tests, we did stacking a "so called" hi end solid state system with all the owner's favourite tweaks, giving us a dead spot-on comparison between JBL and another well known marque's speakers...and then driving them with a mix of music from his system, my own design of amp (which took about 5yrs to decide on the design!) and an ancient but well restored pair of USA made monoblocks from 1947.
There was no doubt in anyone's mind at the end of our marathon listening sessions, that the design I had come up with sounded qualitively much better than the others and made all this digital material sound "good" even when the recording and bit rates were far from optimal.
(FYI we also used some old AKM convertors in the studio grade DAC, compared with commercial DACs with some "tuby plugins/options, and even with a "tube sound" preamp internally equipped with twin triodes...)

Valve amps "can" make this harsh digital sound a quantum leap better and less tiring on the ears, but you don't start by making statements from pure prejudice.

Is there any trope you missed?

 

[horias2000]

"After building several solid state amplifiers, I stared reading about vacuum tube amplifiers. I already knew that the technology is really outdated and no matter how much we try with tubes, transistors will always measure and sound better."

I am sorry you are wrong.
I must also suspect you did all your measurements into a dummy load (a resistor).
Very different things happen once connected to real loudspeakers (especially ESLs!)

We have done many double blind tests, including evaluating whether or not a "hi end" shop owner could tell the difference between hi rate mp3, low rate mp3 and uncompressed 24 bit wav.
(what was so suprising from the so called "golden ears boys" is they preferred the mp3 version, with all those masking artefacts, and could scarcely tell whether or not a DSP was in use to filter off some nasty room resonances)..

I therefore very strongly question whether or not you have the ability to hear the difference between the technologies you claim to, and in reality the above statement was made out of a mix of ignorance or contempt for a subject you knew little about.
Whilst I do accept (having tested loads of them!) that the majority of mass produced valve and sand based amps leave a lot to be desired I simply have to rely on the results of a week of (mostly blind) tests, we did stacking a "so called" hi end solid state system with all the owner's favourite tweaks, giving us a dead spot-on comparison between JBL and another well known marque's speakers...and then driving them with a mix of music from his system, my own design of amp (which took about 5yrs to decide on the design!) and an ancient but well restored pair of USA made monoblocks from 1947.
There was no doubt in anyone's mind at the end of our marathon listening sessions, that the design I had come up with sounded qualitively much better than the others and made all this digital material sound "good" even when the recording and bit rates were far from optimal.
(FYI we also used some old AKM convertors in the studio grade DAC, compared with commercial DACs with some "tuby plugins/options, and even with a "tube sound" preamp internally equipped with twin triodes...)

Valve amps "can" make this harsh digital sound a quantum leap better and less tiring on the ears, but you don't start by making statements from pure prejudice.

Well I might be partially wrong but I do not think I'm wrong in saying that solid state amps (and electronics in general) measure considerably better than tube amps. The proof is in this thread and the thread with the solid state amp I built. The measurements are significantly better for the solid state amp. That being said, I agree with you that the difference is not audible, at least not for me. The amp is now at a friend and he says he can tell the difference and he likes the tube better. He also has a solid state amp I designed and built.
Yes, most of the measurements were performed using resistive loads (wirewound power resistors) but I also did tests using the speakers I have and I did not notice any major differences.
I'm with you when you say that we can't really hear the differences between digital formats. I did these tests myself and there was absolutely no chance to hear any difference between CD audio (16bit - 44.1kHz) and any HD audio formats. If we use the same source file so that we do not have any mastering differences between the files I also believe that it is very hard, if not impossible, to tell the differences. In a blind test that is. That is why I do not believe cables make any differences in a hi-fi system. I can't hear any differences and I can't measure any as well. And there is nobody on this planet that measures any differences between cables (power, network, usb, speaker or interconnect).
In the end I was very pleasantly surprised by the tube amp's sound. I like it even though I can't really hear a difference between it and the solid state amp I built.

 

[horias2000]

That seems like a large spread. Perhaps the UL tapping or anode-anode load impedance isn't optimal for UL with that tube?

Yes, that might be the case indeed. Nevertheless, I can't really see any benefits in trying to make UL mode better. As far as I know, UL mode will yield more power. Having around 2x 20W is more than enough for me. If I wanted more power I would need to increase the supply voltage as I now have a B+ of around 360Vdc. I would also have to use different output transformers with lower primary impedance. The ones I use have 6600ohm primary impedance and as far as I can tell, KT88 will work well with 4000ohm or below. So using a 4000ohm output transformer might yield better THD in UL mode. I might try this at some point.

 

[Eulipian]

Beautiful build very clean looking and a VU meter for DC to boot. How analog can it get? I hope you enjoyed this creative work and that that howlers in the trees can just take a breath and congratulate you on making something wonderful that you enjoyed thrice--designing, building, and listening to. May your tetrodes all remain kinkless.

 

[restorer-john]

You mentioned "pesky ground loops".

This mounting of the toroidal with what appears to be a bolt through to the chassis, with your safety ground wire attached could potentially create a large semi-shorted turn back via the mains earths and RCA outers.

 

[horias2000]

You mentioned "pesky ground loops".

This mounting of the toroidal with what appears to be a bolt through to the chassis, with your safety ground wire attached could potentially create a large semi-shorted turn back via the mains earths and RCA outers.

View attachment 261279

Good point! I don't think it is an issue as the ground wire that goes to that bolt is the primary to secondary screen that is inside the power transformer. So it's not the main EARTH connection for the chassis. The power transformer has this primary to secondary screen inside in order to minimize the parasitic capacitance between primary and secondary. So it just sits between the two windings and is connected to the chassis in one point. It should not create a short of the transformers but I will have another look at this. Thanks!