Hello there,
I have built my first amp these last weeks. It is base on 6SL7-2A3 (from jelabs design, with changes)
Learning from my mistakes, the amp now sounds very good (details, sweetness, harmonic richness, dynamics...) but...
...I had before that a class d amp (trends audio ta10.1) and the only thing i miss the former amp is its speed and transients. I find the tube amp slightly offbeat.
I thought SE DHT could also be renowned also for their speed.
From your experience, what would the key factor in SE DHT design concerning speed, before i go on buying (expensive) stuff ?
- cathode caps quality/value, input or output stages
- psu caps quality/value, first cap or following ones
- input/driver capability (mA) or swing (volts) or Zout (kohm)
- the tubes themselves, especially rectifier (i have 5y3 and gz34)
- signal cap
- other
In my experience, I think everything parallel (ie decoupling caps) creates trouble in music unity (time constants)
What improved speed once : cathode caps removed from 6sl7 but sound became gritty (impedance matching problem with high Zout 6sl7 ?)
My amp (attached):
every connection made with screw connectors to change everything easily until final design.
SOVTEK 2a3 (only 80% on plates because hard mistakes), BRIMAR NOS 6SL7 (and other brands), JAN-NOS-5Y3 or JJ-GZ34
i also tried 6sn7 but prefered 6sl7
I have built my first amp these last weeks. It is base on 6SL7-2A3 (from jelabs design, with changes)
Learning from my mistakes, the amp now sounds very good (details, sweetness, harmonic richness, dynamics...) but...
...I had before that a class d amp (trends audio ta10.1) and the only thing i miss the former amp is its speed and transients. I find the tube amp slightly offbeat.
I thought SE DHT could also be renowned also for their speed.
From your experience, what would the key factor in SE DHT design concerning speed, before i go on buying (expensive) stuff ?
- cathode caps quality/value, input or output stages
- psu caps quality/value, first cap or following ones
- input/driver capability (mA) or swing (volts) or Zout (kohm)
- the tubes themselves, especially rectifier (i have 5y3 and gz34)
- signal cap
- other
In my experience, I think everything parallel (ie decoupling caps) creates trouble in music unity (time constants)
What improved speed once : cathode caps removed from 6sl7 but sound became gritty (impedance matching problem with high Zout 6sl7 ?)
My amp (attached):
every connection made with screw connectors to change everything easily until final design.
SOVTEK 2a3 (only 80% on plates because hard mistakes), BRIMAR NOS 6SL7 (and other brands), JAN-NOS-5Y3 or JJ-GZ34
i also tried 6sn7 but prefered 6sl7
Attachments
What do you mean by "speed"? Bandwidth?
edit: A quick calculation shows that the Miller rolloff and the transformer rolloff won't be too far apart. That's not a good thing. You want to lower the driving impedance. Also, the driver stage will be panting trying to swing the volts to drive the output stage. I'd expect a lot of distortion.
edit: A quick calculation shows that the Miller rolloff and the transformer rolloff won't be too far apart. That's not a good thing. You want to lower the driving impedance. Also, the driver stage will be panting trying to swing the volts to drive the output stage. I'd expect a lot of distortion.
I don't mean a technical aspect (bandwidth...)
I just mean the musical fact of being with the beat, the contrary of a slow/dull perception of music.
I could say toe-tapping...
when I try to follow a rythm, i perceive the amp always being too late
eric
I just mean the musical fact of being with the beat, the contrary of a slow/dull perception of music.
I could say toe-tapping...
when I try to follow a rythm, i perceive the amp always being too late
eric
mahleriana said:
IMHO it is the overall design. It is simple and works but you cannot ask more than that. 🙂
Careful matching with speakers could change a lot your sound. However this is not a cheap way to go, usually.
Maybe you can get a little improvement going to 100 uF for the cap close to the opt and increasing the cathode bypass caps.
Don't use old caps. Use new, freshly made capacitors. You don't need to buy exotic stuff.
In my experience good chinese 2A3's, like Golden Dragon or Billington brands, sound better than Sovtek.
If you wish to improve a lot your amp you need to go for another design.
I would start from the Audio Note Neiro or something like this: http://www.positive-feedback.com/Issue1/kismet.htm
Cheers,
45
mahleriana said:
My recipe includes at least 10mA of quiescent current in the preceding driver stage and fixed bias in the output stage. No electrolytics anywhere in the signal path.
Note that most large value motor run caps sound very "romantic" in the highs due to an appreciable amount of ESL/ESR - need to get some help using the Western Electric connection with higher quality film caps. (Can hum buck too. Search my older posts on this subject.) An unbypassed MRC is still better IMO than any electrolytic I have heard except a now virtually unobtainium Black Gate or Elna..
My current design uses 6SN7 based srpp driver running at 10mA and fixed bias on a 300B.. Some people might have an issue with this, but try it before you knock it. Note you need an additional gain stage ahead of the srpp unless you have a linestage with appreciable gain in which case with the 2A3 it would be optional..
Another very good option might be a choke or CCS loaded D3A (triode connected) or 5842/417A running at 15 - 20mA.. (More than sufficient swing to drive a 2A3.)
Note that the 6SL7 rp and rl conspires to produce a driver stage source impedance that is too high to effectively drive the miller capacitance of your 2A3.. (Even if such were not the case there is not enough current available to do the job.)
Look at my website for a design that uses the 6SL7 to drive the 45, it has a fairly low output impedance and does a nice job with the 45 and 2A3..
Excessive even order distortion and probably too low damping for your speakers. Kevin's recommendations seem spot-on.
I thought SE DHT could also be renowned also for their speed.>>
That's what I've found. But you don't have a DHT amp - you have an amp where the majority of the amplification is done by the 6SL7, which I find to have a rather thick romantic sound.
If you want the DHT sound build an all DHT amp.
andy
That's what I've found. But you don't have a DHT amp - you have an amp where the majority of the amplification is done by the 6SL7, which I find to have a rather thick romantic sound.
If you want the DHT sound build an all DHT amp.
andy
I think it is not a problem from the speakers : TQWT designs of different fostex single drivers.
I sometimes read than simple anode follower stages are more musical than SRPP, mu followers etc (more "hifi") ...but i never other than anode-f.
Anyway my philosophy is often the simpler the better... Fi X or Decware Zen made me desire to build my own amp... I intend to go Direct Coupled when i will be more sure of my design.
Anyway the slow aspect is just slight, but enough to bother me.
perhaps a separate anode loaded input+driver ?
eric
I sometimes read than simple anode follower stages are more musical than SRPP, mu followers etc (more "hifi") ...but i never other than anode-f.
Anyway my philosophy is often the simpler the better... Fi X or Decware Zen made me desire to build my own amp... I intend to go Direct Coupled when i will be more sure of my design.
Anyway the slow aspect is just slight, but enough to bother me.
perhaps a separate anode loaded input+driver ?
eric
You need to change what you read.😀 People make all sorts of silly unsupported pronouncements. That doesn't make them true.
mahleriana said:
When it comes to the grid of a power tube the simplest solution is not always the best choice, especially if you have a limited budget.
I think that the weakness in your amp is not the SE driver itself but the RC-coupled 6SL7 and the self-bias for the 2A3.
Switching to a Loftin-White circuit could improve your amp.
However, IMO, you are not taking out the best from the 2A3, yet.
Cheers,
45
I know, I know... the problem being there are tens of typologies, hundreds of designs and thousand of factors...and no SE amp around me to listen to but mine.
Instinctively, i would like to keep away from any sort of feedback so...
I should perhaps for the moment limit the question to : how to optimize regarding speed a 2 or 3 stage anode loaded amplifier, because i understand it more than the others.
If it does not work, I will then go to another design.
I intended to try another tube to replace 6sl7 : 6922/6dj8/6n1p kind or perhaps 5687 or 5842. Audion use for example 6922>5687
They seem to have much lower Rp, sufficient current and perhaps enough mu.
your opinion ?
I think replace electrolytics by pp would not be cheap... i could try cathode bypass of 1st stage from 100uF to 470 or 1000 staying electrolytic low voltage = cheap try
Instinctively, i would like to keep away from any sort of feedback so...
I should perhaps for the moment limit the question to : how to optimize regarding speed a 2 or 3 stage anode loaded amplifier, because i understand it more than the others.
If it does not work, I will then go to another design.
I intended to try another tube to replace 6sl7 : 6922/6dj8/6n1p kind or perhaps 5687 or 5842. Audion use for example 6922>5687
They seem to have much lower Rp, sufficient current and perhaps enough mu.
your opinion ?
I think replace electrolytics by pp would not be cheap... i could try cathode bypass of 1st stage from 100uF to 470 or 1000 staying electrolytic low voltage = cheap try
for the loftin white i am not sure i will have enough B+ from my 300-0-300 transformer
instead of self bias : what would you recommend ?
instead of self bias : what would you recommend ?
kevinkr said:
Motor starts certainly, but motor runs? ASC, for example, specs the same DF for their metalized PP oilers as for the regular tape or dipped versions. The inductance of those wonderfully neat and fastidiously power supply harnesses in many home builds swamps the ESL of any decent modern cap in my experience. And what's so wrong with romance?
I don't know if one of the previous posts was asking me for more info, but if it was then the valves I would suggest are any of:
Input: ECC40, 01A
Driver: 71A, 31, 26.
These are easy to implement, since their filaments aren;'t so greedy (except the 26).
You will notice that the ECC40 isn't a DHT. It's about the only indirectly heated tube (with the 2C22) that has for me some of the delicacy I find with DHTs. It's useful if you need more gain.
I usually build balanced circuits and I've been having good luck with the 1J6G, but that's a double DHT with shared cathode.
andy
Input: ECC40, 01A
Driver: 71A, 31, 26.
These are easy to implement, since their filaments aren;'t so greedy (except the 26).
You will notice that the ECC40 isn't a DHT. It's about the only indirectly heated tube (with the 2C22) that has for me some of the delicacy I find with DHTs. It's useful if you need more gain.
I usually build balanced circuits and I've been having good luck with the 1J6G, but that's a double DHT with shared cathode.
andy
rdf said:
I understand what you are saying and I like and use ASC motor runs myself (avoid motor starts like the plague) but electrically they are not as good as they are cracked up to be - that is not to say they aren't a heck of a lot better than any comparable value electrolytic.
Measuring the self resonant frequency of the cap should give you a good sense of the parasitic reactances and from the Q of that resonance other losses like the ESR/DF. (I do remember that the resonant frequency was lower than I expected - under 20kHz, implying at least 1uH of esl. You're looking for a series resonant notch, just stick a 100 ohm - 1K resistor in series with the cap and sweep it with a function generator.)
I just have not quite figured out how to do this very accurately - I will say the 100uF ASC motor runs I purchased on eBay had pretty low DA, but from an AC perspective had appreciable audio across them from a couple of kHz on up (and increasing with frequency) unfortunately I did not do precise measurements or document any of this so please take this as merely anecdotal.
I was really more interested in fixing the problem than investigating the cause - so I went to a hum canceling configuration using some 22uF Solen fast caps which improved the situation greatly. (What I refer to as the "Western Electric Connection." 100uF cathode to ground, 22uF cathode to B+)
I suspect smaller motor run caps will have significantly better high frequency performance than the huge 100uF cans I used. (They're good in psu too, too bad I can't get any more.)
Mahleriana, you could go for a back-bias!
It is a simple and cheap fixed bias solution which doesn't require any additional supply!!🙂
If I am not wrong it was implemented in one of the old Brooks amplifiers. It works only with class A amplifiers because one of the requirements is that the total DC current of the amplifier has to be costant for any modulation level.
In the real world this is never true 100%, however it works much better than cathode bias.
Tomorrow I will load a schematic if you like.
The only downside will be that you will lose about 45V (i.e. the 2A3 bias voltage) at the front-end as well.
However if you switch to 6922 it is not a problem at all.
Cheers,
45
It is a simple and cheap fixed bias solution which doesn't require any additional supply!!🙂
If I am not wrong it was implemented in one of the old Brooks amplifiers. It works only with class A amplifiers because one of the requirements is that the total DC current of the amplifier has to be costant for any modulation level.
In the real world this is never true 100%, however it works much better than cathode bias.
Tomorrow I will load a schematic if you like.
The only downside will be that you will lose about 45V (i.e. the 2A3 bias voltage) at the front-end as well.
However if you switch to 6922 it is not a problem at all.
Cheers,
45
- Status
- Not open for further replies.