Pros and cons of this?

Hi all...

I have some thoughts of building a HiFi amp for use at home and after looking through the Internet I have seen various schematics and I also have understood that a pretty high THD is... something preferred.

SRPP's and 300B triodes are highly sought for.

That put aside for the moment I came across something called SIPP (self inverting push-pull) and the designer use a LM317L regulator as a constant current source and I wonder what I could use instead of this? Could it work with a fixed bias or even tap some, I don't know, 40 - 50V from the secondary AC as in some guitar amps?

Hope the question is understandable

Best regards

/John

johnheath said:

Hi all...

I have some thoughts of building a HiFi amp for use at home and after looking through the Internet I have seen various schematics and I also have understood that a pretty high THD is... something preferred.

Click to expand...

Maybe - it depends on the harmonic properties of the distortion of course.
However, it is by definition not 'High Fidelity'  ;D

Newmarket said:

Maybe - it depends on the harmonic properties of the distortion of course.
However, it is by definition not 'High Fidelity'  ;D

Click to expand...

Yes, so I understand. But tell me, Can you say anything about the circuit with the KT88's? Could that regulator be substituted with fixed bias or be biased from a rectified tap from the secondary AC?

Best regards

/John

Hi,

Could that regulator be substituted with fixed bias or be biased from a rectified tap from the secondary AC?

Click to expand...

The output stage is a differential ultra linear output stage so the ccs will help the circuit about distortion.
You could substitute the ccs for a resistor and make it a more traditional longtailed pair but you may loose a bit of accuracy as an outcome.

May i ask why you want to get rid off the ccs? Are you worried about noise? As far as i remember used as a CCS this kind of regulator are ok, and being a CCS it help isolate from ground noise by its own nature.

You coud make your own ccs using a cascode mosfet or bjt instead of the lm317. Given price and simplicity they have great performance/price ratio.

I have some thoughts of building a HiFi amp for use at home and after looking through the Internet I have seen various schematics and I also have understood that a pretty high THD is... something preferred.

Click to expand...

Well this circuit may not have a high distortion: the differential nature of output stage will tend to cancel 2nd order harmonics ( and the ccs will make this even more true as well as isolate from ground noise as already pointed), being ultra linear it does have degeneration applied too (local feedback through grid2) so the distortion should be relatively low.
The srpp stage is nested in another global feedback loop so distortion will be even lower for even/odd orders harmonics too.

Maybe another choice of first stage toplogy could have been wiser, SRPP being more a power amplifier than a pure voltage gain amplifier (there is a great explanation about that in Merlin Blencowe paper about SRPP you can find at 'valvewizard' site). But the increased current given by this topology (srpp)  may be required to drive the kt88 grid?

From my own experiment SRPP does have a 'nice and pleasing' sound (euphonic) without feedback when used with low/medium mu tube (something like 6sn7 around 20 or lower for mu, or 6n6p/5687 family if you don't fear the high heater current), enough current through it (8ma for 6sn7, 12au7,... more like 10/12ma for 6n6p/5687) and with a load impedance greater than 50/60k.

Once nested in feedback loop it can have low harmonic distortion ( Ian's gainstage (ruffrecord) is a nice and effective example).

Hope this help. 

Krivium's comments are well above what I can contribute to the tubey side of things Good Luck

Maybe - it depends on the harmonic properties of the distortion of course.

Click to expand...

I fully agree.

Krivium's comments are well above what I can contribute to the tubey side of things

Click to expand...

Maybe, but you may contribute to the overall stage topology as well, which is imho as or more important than the components used (tube, fet,bjt... all have theyr pro and cons but the topology where they are used is more relevant soundwise imho... but i'm more toward tubes if i have choice!).

SRPP's and 300B triodes are highly sought for.

Click to expand...

Yes but used in single ended topology (for the 300b which is a power triode) and more often than not without feedback ( for good and bad reason imho).

The circuit you plan to use use a pushpull output stage with pentode and local feedback (and 2 instance of it: the ultralinear grid and the 1r resistors in the cathode... and the pentode is used as tetrode for what i understand from the schematic), and a single ended pushpull (Shunt Regulated Push Pull) nested in a feedback loop.

This is very different from what you expect at first.
This is not to say this circuit won't sound good neither that feedback is wrong.

I have come across this circuit before. It is a simple way of getting the output stage to be its own phase splitter. I think there is even a name for it.  Apparently it has some disadvantages over a regular phase splitter but I cannot remember what. It virtue is its simplicity. In all other respects it is pretty much the same as any push pull output stage.

Cheers

Ian

KrIVIUM2323 said:

Hi,
The output stage is a differential ultra linear output stage so the ccs will help the circuit about distortion.
You could substitute the ccs for a resistor and make it a more traditional longtailed pair but you may loose a bit of accuracy as an outcome.

May i ask why you want to get rid off the ccs? Are you worried about noise? As far as i remember used as a CCS this kind of regulator are ok, and being a CCS it help isolate from ground noise by its own nature.

You coud make your own ccs using a cascode mosfet or bjt instead of the lm317. Given price and simplicity they have great performance/price ratio.

Click to expand...

Thank you sir...

The reason I would like to get rid of the regulator is simply because I would prefer not to use regulators or transistors and such at all... it is surely a very good solution to use them but I just would like not to.

About the topolgy in the first step I might have other solutions... the topology is not my concern... it is all about what I can use instead of the regulator and/or if it is possible to use another setup to bias the output tubes in this fashion from a SRPP or other first stage?

And also I am not about to copy the schematic... just have some questions about the output topology

Best regards

/John

ruffrecords said:

I have come across this circuit before. It is a simple way of getting the output stage to be its own phase splitter. I think there is even a name for it.  Apparently it has some disadvantages over a regular phase splitter but I cannot remember what. It virtue is its simplicity. In all other respects it is pretty much the same as any push pull output stage.

Cheers

Ian

Click to expand...

Thanks Ian

This is my main concern... I was thinking of building this "HiFi amp" with a AC long tail pair as a phase splitter into a push-pull output tubes... but I came across this schematic and the designer seems to have a couple of variants of it and they seem to be popular, but I am just trying to understand the ... pros and cons.

Perhaps first gain stage into a long tail pair will produce too much gain for "HiFi" so careful NFB must be considered?

Best regards

/John

Hi,
I don t see any reason you cannot bias the output stage with a more 'conventional' way.
In fact the Hafler/Keroes ultra linear amplifier does use resistor to bias it s output stage (cap bypassed).

The ccs in the schematic just 'improve' the circuit in case you use it as being it s own phase splitter ( well i would say it does improve the circuit anyway, and it insure classA operation which may not be the case if you bias it differently).

You can see what to expect about distortion when a differential stage is used without phase splitter in preceding stage in  erno boberly's article 'jfet the final frontier' where he give numbers about it.

You can make ccs using tube (and tube cascode even better) but this complicate things a bit. You may need negative psu for this.

Anyway, make a search for the Hafler ultra linear amplifier to see article and schematic, you ll have analysis and why they decided to use the ultra linear topology ( it is very low distortion, between triode and pentode behavior, lower output z... used by mcintosh, Ear,...).

My comment about topology was just because you seem to look for triode sound in your reference and make choice of something really different. That said the incarnation of this kind of topology (ultalinear) i ve heard are quite good sounding to me ( mcintosh amplifier, Ear 660. Never heard an Hafler but would like to).

The self-split needs twice the drive voltage and cant NOT go into class AB.

It does cancel DC in the OT.

It makes sense for a $5 amp. I never saw it at KT88 level. It can work. There are MANY other plans. Clearly you have not really begun the endless hunt for tube hi-fi.

KrIVIUM2323 said:

Hi,
I don t see any reason you cannot bias the output stage with a more 'conventional' way.
In fact the Hafler/Keroes ultra linear amplifier does use resistor to bias it s output stage (cap bypassed).

The ccs in the schematic just 'improve' the circuit in case you use it as being it s own phase splitter ( well i would say it does improve the circuit anyway, and it insure classA operation which may not be the case if you bias it differently).

You can see what to expect about distortion when a differential stage is used without phase splitter in preceding stage in  erno boberly's article 'jfet the final frontier' where he give numbers about it.

You can make ccs using tube (and tube cascode even better) but this complicate things a bit. You may need negative psu for this.

Anyway, make a search for the Hafler ultra linear amplifier to see article and schematic, you ll have analysis and why they decided to use the ultra linear topology ( it is very low distortion, between triode and pentode behavior, lower output z... used by mcintosh, Ear,...).

My comment about topology was just because you seem to look for triode sound in your reference and make choice of something really different. That said the incarnation of this kind of topology (ultalinear) i ve heard are quite good sounding to me ( mcintosh amplifier, Ear 660. Never heard an Hafler but would like to).

Click to expand...

Thank you sir...

I surely will have a look at those Hafler's and their schematics to got more influences and information. I have been looking at McIntosh before but they seem to use their own output transformers that are pretty advanced and (probably) very expensive.

Best regards

/John

PRR said:

The self-split needs twice the drive voltage and cant NOT go into class AB.

It does cancel DC in the OT.

It makes sense for a $5 amp. I never saw it at KT88 level. It can work. There are MANY other plans. Clearly you have not really begun the endless hunt for tube hi-fi.

Click to expand...

Thank you Sir...

These are facts that I was looking for... among many other thing as well of course

Yes, the designer offers kit's and schematics for DIY, and as you say I am far from having seen much of the hi fi world.

To be honest I am not that picky about it, but as I wrote... I will use this amp for personal use at home... you know, just for listening to old records and perhaps some CD's as well

Still searching and learning

Best regards

/John

Macintosh uses some pretty weird transformers, and some are tri-filar wound so you have to use triple build enamel wire,  or Kapton coated if you want to be like Doc Hoyer.

what about the Williamson circuit with the Peerless 15036 monster outputs?
probably not realistic so scratch that,

build a normal long tail inverter but stick a choke in the cathode if you want to be different.

Thank you sir

Yes, I have been investigating those McIntosh before and even highlighted them with a couple of questions here on the forum a couple of years ago... I felt that I had to leave them aside because of their circuits with their advanced transformers.

I'll have a look at the Williamsons

And I also believe that a regular long tail pair as a phase splitter probably would be the easiest way for me. I also took a quick glance at the Dynaco Tube amps and there are a couple of interesting features there.

Best regards

/John