Following the thread I started on the "Opinions on Walton Audio 300B Design" which started a very lively debate, I have decided to pursue a different design.
I will be building this design by SJS Electroacoustics
I welcome any comments on the design. I'm particularly interested in the Power Supply as I'd like to change the 470uF capacitors to something smaller - as I want to use a quality unit and many do not go to these values.
Please keep all discussions on the topic of design and construction of this amplifier.
Many Thanks
Ralf
I will be building this design by SJS Electroacoustics
An externally hosted image should be here but it was not working when we last tested it.
I welcome any comments on the design. I'm particularly interested in the Power Supply as I'd like to change the 470uF capacitors to something smaller - as I want to use a quality unit and many do not go to these values.
Please keep all discussions on the topic of design and construction of this amplifier.
Many Thanks
Ralf
Konnichiwa,
Main one, experiment with lower current for the Driver valve (10 - 12mA rather than the > 20mA used. The result is somewhat higher 2nd Harmonics and MUCH lower upper harmonics which when mixed with the output stage distortion result inmuch lower high order products and somewhat lowered 2nd harmonics.
Seconly, note the 100uF Electrolytic capacitor in the fixed bias supply is actuallu IN THE SIGNALCURRENT loop. It acts as coupling capacitor in the circuit, so you retain a coupling capacitor, though not in the usual place and hence invisible, PLUS a much lower quality one.
Also, you have a massive Electrolytic in the Driver Cathode, not good at all. You may wish to experiemnt with more modest values of Film Capacitors (start with 100uF). Thgis can with the primary inductance of the Interstage Transformer lead to a peakerd up lower LF response which with the right "tuning" can extend the LF response of the Amplifier to something ebtter than possible with just the output transformer.
Well, you could use one 50H Choke and one 470uF Capacitor. Or you could use two LC Cells, with easily obtained 10H/100R Chokes (make sure the first one is suited to choke input and the second one offers as high a resonance as possible) and 50uF Oil/Paper or Film Capacitors.
In the first case you get around 15mV hum on the +B supply and around 0.65mV Output Hum (assuming NO other hum sources). In the second case you get 2.5mV hum on the +B supply and 0.1mV Output Hum (assuming no other sources).
The downside is that the supply with the two 10H/50uF cells has a slightly poorer load step response, which may be audible. If you inchrease the seond choke to 20H and use two 50uF Capacitors things improve notably and hum is even lower (0.2mV on +B line).
For the 300B Heater Supply make sure the transformer has an electrostatic screen and I would recommend using a suitable Choke in the filter circuit, instead of the Resistor. Maybe 100mH/1.5A/1...2 Ohm should be easily doable using normal winding machines and Iron.
Such a chokje has an impedance of 63 Ohm @ 100Hz, compared to 1 Ohm @ 100Hz for the 1 Ohm Resistor. This gives around 30db lower noise on the heater (which is also the cathode) than the simple RC circuit. Using the proposed LC filter the noise on the heater is around 0.4mV with the RC filter around 15mV and has a waveshape that is quite rich in higher harmonics, which might cause an audible "greying out" of the sound.
Anyway that would be my suggestions that exclude fundamental re-designs to eliminate some of the subideal approaches taken.
Sayonara
DVDHack said:
Main one, experiment with lower current for the Driver valve (10 - 12mA rather than the > 20mA used. The result is somewhat higher 2nd Harmonics and MUCH lower upper harmonics which when mixed with the output stage distortion result inmuch lower high order products and somewhat lowered 2nd harmonics.
Seconly, note the 100uF Electrolytic capacitor in the fixed bias supply is actuallu IN THE SIGNALCURRENT loop. It acts as coupling capacitor in the circuit, so you retain a coupling capacitor, though not in the usual place and hence invisible, PLUS a much lower quality one.
Also, you have a massive Electrolytic in the Driver Cathode, not good at all. You may wish to experiemnt with more modest values of Film Capacitors (start with 100uF). Thgis can with the primary inductance of the Interstage Transformer lead to a peakerd up lower LF response which with the right "tuning" can extend the LF response of the Amplifier to something ebtter than possible with just the output transformer.
DVDHack said:
Well, you could use one 50H Choke and one 470uF Capacitor. Or you could use two LC Cells, with easily obtained 10H/100R Chokes (make sure the first one is suited to choke input and the second one offers as high a resonance as possible) and 50uF Oil/Paper or Film Capacitors.
In the first case you get around 15mV hum on the +B supply and around 0.65mV Output Hum (assuming NO other hum sources). In the second case you get 2.5mV hum on the +B supply and 0.1mV Output Hum (assuming no other sources).
The downside is that the supply with the two 10H/50uF cells has a slightly poorer load step response, which may be audible. If you inchrease the seond choke to 20H and use two 50uF Capacitors things improve notably and hum is even lower (0.2mV on +B line).
For the 300B Heater Supply make sure the transformer has an electrostatic screen and I would recommend using a suitable Choke in the filter circuit, instead of the Resistor. Maybe 100mH/1.5A/1...2 Ohm should be easily doable using normal winding machines and Iron.
Such a chokje has an impedance of 63 Ohm @ 100Hz, compared to 1 Ohm @ 100Hz for the 1 Ohm Resistor. This gives around 30db lower noise on the heater (which is also the cathode) than the simple RC circuit. Using the proposed LC filter the noise on the heater is around 0.4mV with the RC filter around 15mV and has a waveshape that is quite rich in higher harmonics, which might cause an audible "greying out" of the sound.
Anyway that would be my suggestions that exclude fundamental re-designs to eliminate some of the subideal approaches taken.
Sayonara
Does that 100uF Cap and Resistor network need to be there at all - is it just setting the Voltage?
I want to change those 470uF caps to 220uF, can I do that? Going to try and model this stuff in that PS Application.
That 2200uF Cap is there to avoid the Bottom end roling off I think, so I'm happy to change it to a smaller bypass value - I'll determine the size once I decide if I'm running a Sub Woofer on this system, thinking I'll get away with less efficient speakers by eliminating Bass load and get beter sonic performance.
I'll probably add a switch to allow me to change the bottom end response by putting additional capacitance in place - then its suitable with non sub situation as well.
I want to change those 470uF caps to 220uF, can I do that? Going to try and model this stuff in that PS Application.
That 2200uF Cap is there to avoid the Bottom end roling off I think, so I'm happy to change it to a smaller bypass value - I'll determine the size once I decide if I'm running a Sub Woofer on this system, thinking I'll get away with less efficient speakers by eliminating Bass load and get beter sonic performance.
I'll probably add a switch to allow me to change the bottom end response by putting additional capacitance in place - then its suitable with non sub situation as well.
hi,
it is there to make sure you get accurate voltage to bias the output tube.
cheers.
DVDHack said:
it is there to make sure you get accurate voltage to bias the output tube.
better. i simulated it in the past and i was getting weird results, so i dumped that section and made a new one. however, my IT has not arrived. i'm stuck.
i feel the same way. in my 5842 dc'ed to 245 i also have a high value, but i don't have a scope yet to try to figure out how low i can get it to.
cheers.
You may need those 470uF caps on the power supply, due to the choke input there is a lot of ripple. But 470uF/450V is a pretty common value these days due to computer power supplies. Also make sure to use very, very good chokes otherwise they will buzz mechanically. Most chokes do when you put ripply DC through them.
In my opinion, choke input is wasted on this design. Single ended amps have fairly constant current draw, so the better regulation from a choke-capacitor (L-C) input filter is not really noticeable over plain CLC filter. And you would need at least 3 chokes and 3 rectifier tubes, and a custom power transformer! Ouch.
The 2200 6.3V bypass cap looks huge to me, but it might be due to the I.T. i suppose... I use about 470uF on the first stage in my R-C coupled design and thats huge overkill.
And why the 437A? As far as I know its a radio frequency tube a bit like a 12AT7. Oh hang on, its made by Western Electric Must be amazing
Heh don't mean to rain on the parade but why not try a more conventional design like the Flesh and Blood for your first amp?
In my opinion, choke input is wasted on this design. Single ended amps have fairly constant current draw, so the better regulation from a choke-capacitor (L-C) input filter is not really noticeable over plain CLC filter. And you would need at least 3 chokes and 3 rectifier tubes, and a custom power transformer! Ouch.
The 2200 6.3V bypass cap looks huge to me, but it might be due to the I.T. i suppose... I use about 470uF on the first stage in my R-C coupled design and thats huge overkill.
And why the 437A? As far as I know its a radio frequency tube a bit like a 12AT7. Oh hang on, its made by Western Electric
Must be amazing Heh don't mean to rain on the parade but why not try a more conventional design like the Flesh and Blood for your first amp?
Konnichiwa,
The resistor network sets the bias voltage, the Capacitor closes the current loop for audio signals. Both are needed to have the circuit function as intended, BUT given the DC resistances around it the capacitor could be a much lower value Film Type, I'd try 10uF film of good quality.
Yes, but hum is going up. Plus there are few readily available
non-electrolytic of that kind of value anyway and you really do not want ANY electrlytic capacitors in signal current loops if avoidable.
Not DIRECTLY. It reduces the increase in aprarent anode impedance. If the value is too small the effective anode impedance of the Valve rises with lowering frequency, which MAY lead to LF rolloff in conjunction with the load.
If you operate such a stage with an unloaded interstage transformer as load, the cathode capacitor, cathode resistor, psu capacitor and valve form a damped resonant circuit, whose resonance usually falls below 10Hz and can be used to boost
the LF response a little by correct tuning extending the LF response.
For that we "preemphasise" the drive signal to the 300B to offset some of the LF rolloff caused by the (loaded seconadary and thus usually well damped) Output Transformer. If the LF lift ends up too much, a "damping resistor" of a few 10's of KOhm can be used to finetune the degree of lift, the cathode capacitor tunes the frequency of the lift with a given IT.
Ideally we want a non-electrolytic capacitor in the cathode, of high quality. As these are rarely found in large enough values (exceptions are the Rubycon SWR series which includes a nice sounding 100uF/35V Mylar Cap), so electrolytics are used. Electrolytic capacitors have rather high levels of various distortions, these are best minimised by using ovsezized Units, which sadly again limits the choice rather more than one likes.
I think Film Cap's like the mentioned Rubycon are a better choice.
Actually, a lower value cathode bypass cap in the circuit you intend to use will extend the LF and possibly even lead to a bass boost.
Ripple cleanup is best done using a second LC cell with a choke wound for minimal shunt capacitance.
Only as Electrolytic Capacitors, which are a poor choice for Amplifiers.
Very high transconductance allows very low drive impedance. It can be substituted with a range of similar valves including triode wired pentodes, just look for a high transconductance and > 4W Anode dissipation, plus a Mu > 40.
The F&B sounds okay but not overly exciting (non of the 2-Stage 6SN7 Driver Circuits do to my ears) and our friend does want build his first SE Amp, he wants his ONLY SE Amp....
Sayonara
DVDHack said:
The resistor network sets the bias voltage, the Capacitor closes the current loop for audio signals. Both are needed to have the circuit function as intended, BUT given the DC resistances around it the capacitor could be a much lower value Film Type, I'd try 10uF film of good quality.
DVDHack said:
Yes, but hum is going up. Plus there are few readily available
non-electrolytic of that kind of value anyway and you really do not want ANY electrlytic capacitors in signal current loops if avoidable.
DVDHack said:
Not DIRECTLY. It reduces the increase in aprarent anode impedance. If the value is too small the effective anode impedance of the Valve rises with lowering frequency, which MAY lead to LF rolloff in conjunction with the load.
If you operate such a stage with an unloaded interstage transformer as load, the cathode capacitor, cathode resistor, psu capacitor and valve form a damped resonant circuit, whose resonance usually falls below 10Hz and can be used to boost
the LF response a little by correct tuning extending the LF response.
For that we "preemphasise" the drive signal to the 300B to offset some of the LF rolloff caused by the (loaded seconadary and thus usually well damped) Output Transformer. If the LF lift ends up too much, a "damping resistor" of a few 10's of KOhm can be used to finetune the degree of lift, the cathode capacitor tunes the frequency of the lift with a given IT.
Ideally we want a non-electrolytic capacitor in the cathode, of high quality. As these are rarely found in large enough values (exceptions are the Rubycon SWR series which includes a nice sounding 100uF/35V Mylar Cap), so electrolytics are used. Electrolytic capacitors have rather high levels of various distortions, these are best minimised by using ovsezized Units, which sadly again limits the choice rather more than one likes.
I think Film Cap's like the mentioned Rubycon are a better choice.
DVDHack said:
Actually, a lower value cathode bypass cap in the circuit you intend to use will extend the LF and possibly even lead to a bass boost.
ShiFtY said:
Ripple cleanup is best done using a second LC cell with a choke wound for minimal shunt capacitance.
ShiFtY said:
Only as Electrolytic Capacitors, which are a poor choice for Amplifiers.
ShiFtY said:
Very high transconductance allows very low drive impedance. It can be substituted with a range of similar valves including triode wired pentodes, just look for a high transconductance and > 4W Anode dissipation, plus a Mu > 40.
ShiFtY said:Heh don't mean to rain on the parade but why not try a more conventional design like the Flesh and Blood for your first amp?
The F&B sounds okay but not overly exciting (non of the 2-Stage 6SN7 Driver Circuits do to my ears) and our friend does want build his first SE Amp, he wants his ONLY SE Amp....
Sayonara
As far as the WE 437 is concerned, you'd probably wan't to go for the 6S45P-e instead, which is more readily available, and is a pretty close equivalent. It'll run you something like USD 10 ea.
As for Torsten's suggestion to halve the bias current, I think this depends on the current capacity of the interstage transformer. If necessary, you could use a PP-to-SE interstage transformer, and a rheostat to find the appropriate flux density.
Most people I've seen tend to comment that the 6S45P-e performs ideally at 40mA and above. Granted, these people usually go for a capacitor instead of an interstage, but again, I think that's because not all SE-to-SE interstages like 40mA DC. A couple of the Lundahls do, however.
As for Torsten's suggestion to halve the bias current, I think this depends on the current capacity of the interstage transformer. If necessary, you could use a PP-to-SE interstage transformer, and a rheostat to find the appropriate flux density.
Most people I've seen tend to comment that the 6S45P-e performs ideally at 40mA and above. Granted, these people usually go for a capacitor instead of an interstage, but again, I think that's because not all SE-to-SE interstages like 40mA DC. A couple of the Lundahls do, however.
Konnichiwa,
Actually, the suggestion to lower the Current have more to do with making the Driver stage more linear. Lower current means slightly increased 2nd harmonics (around 6db) but DRASTICALLY DECREASED higher harmonics. The difference in sound is quite profound. Valves like 6S45 or WE437A at somthing like 40mA tend to sound hard and steely when used to drive a 300B and show a quite substantial number of higher harmonics coupled with a fairly substantial levels.
Note that the "low current/high voltage" operation is for use with Inductive or active loads ONLY, not generally for resistively loaded stages.
Sayonara
angel said:
Actually, the suggestion to lower the Current have more to do with making the Driver stage more linear. Lower current means slightly increased 2nd harmonics (around 6db) but DRASTICALLY DECREASED higher harmonics. The difference in sound is quite profound. Valves like 6S45 or WE437A at somthing like 40mA tend to sound hard and steely when used to drive a 300B and show a quite substantial number of higher harmonics coupled with a fairly substantial levels.
Note that the "low current/high voltage" operation is for use with Inductive or active loads ONLY, not generally for resistively loaded stages.
Sayonara
With all due respect to Thorsten I would suggest that you build the thing EXACTLY per schematic. Listen to your amp for a month or three till everything is settled and virtually run in.
And then implement changes...(Thorsten's or others ) that way you learn more and won't have to rely on the opinion of this or that person or flavour of the month part...
For instance I would suggest you try Black Gate's in the PSU...which are as good if not better SOUNDING than film caps. But there are many that won't agree!!! So who is right? You will have to find out what is right for YOU.
Cheers,
Bas
And then implement changes...(Thorsten's or others ) that way you learn more and won't have to rely on the opinion of this or that person or flavour of the month part...
For instance I would suggest you try Black Gate's in the PSU...which are as good if not better SOUNDING than film caps. But there are many that won't agree!!! So who is right? You will have to find out what is right for YOU.
Cheers,
Bas
DVDHack said:
I looked into this point myself, as several fixed bias designs use 50uF or so here and it seems larger than necessary. I think it's like this.
When you turn the power off, both bias voltage and out put stage B+ will run down (pretty obvious really). However, if there is little in the way of capacitance in the bias supply it will drop to zero rapidly, while there is still a lot of B+ voltage there (especially if there are those big 470uF caps in the PSU). High B+ and no bias on the 300Bs will cause an overload current.
It seems the (oversize) 100uF, 50uF etc is used to act as a reservoir and maintain bias for a good few seconds while the B+ drains down.
I'd suggest only going below 50uF if the main PSU is redesigned to use much lower capacitance. I'd suggest this anyway, using CLCLC two pi filtering.
For the main PSU, look into say:
6uF, 10H, 20uF, 10H, 30uF where the chokes should be over rated for
low DC resistance, and the larger caps can be poly/oil motor run types such as GE 97F series. Oh, and model it like mad in PSUD, but also do the pole and Q calculations.
Good luck; it should be a really nice amp.
OK, people mention some issues; but no design is perfect and I think it'll be great.
Konnichiwa,
If you halve the PSU capacitance the hum will go up by a factor 2 or by 6db. As it stands, the amplifier will have around 0.6mV hum on the 8 Ohm with the original supply (assuming no other hum sources). Going to 220uF will push that Hum up to 1.2 - 1.5mV (with a 50Hz supply). If so much hum on the output is tolerable or not you have to decide.
Remember, the supply is in the signal current loop. If you use two LC cells you can use electrolytic capacitors in the first position.
Sayonara
DVDHack said:
If you halve the PSU capacitance the hum will go up by a factor 2 or by 6db. As it stands, the amplifier will have around 0.6mV hum on the 8 Ohm with the original supply (assuming no other hum sources). Going to 220uF will push that Hum up to 1.2 - 1.5mV (with a 50Hz supply). If so much hum on the output is tolerable or not you have to decide.
DVDHack said:
Remember, the supply is in the signal current loop. If you use two LC cells you can use electrolytic capacitors in the first position.
Sayonara
Trouble is I have no idea how much audible hum those ripple values equate to. I realise I can parallel up the Caps but big cost there.
In another thread (layout of this amp) it was suggested I could use a common TFR LC stage for the PSU and then a separate second LC stage for the two channels - still maintaining separation through second LC. What are your thoughts around this?
This should become a good thread for anyone ever contemplating building this design.
Thanks
Ralf
In another thread (layout of this amp) it was suggested I could use a common TFR LC stage for the PSU and then a separate second LC stage for the two channels - still maintaining separation through second LC. What are your thoughts around this?
This should become a good thread for anyone ever contemplating building this design.
Thanks
Ralf
Konnichiwa,
With 100/120Hz the hum is much more audible than 50Hz hum. for reference, 1.4mV would be 66db below 1W (on the 8 Ohm tap and it scales to the other Taps). So, if your speakers have 85db/W/m you get a steady state Hum with an absolute SPL of 19db at 1m Distance, probably inaudible at the listening posiition. If you use 103db/W/m sensitive Beauhorn Virtuoso Speakers (and they are 103db/1W @ 100Hz) the hum level would be 37db at 1m Distance, I personally would find such levels of hum certainly objectionable, I of course cannot speake for you.
My current Amp's are running on speakers that are around 96db/1W/1m @ 100Hz and have around 0.2...0.5mV 100Hz from the use of AC heaters (they are 45 Push-Pull). I find this objectionable only if listening very quiet late at night.
You still keep the grounds common. This may cause more problems than you think. Also, suddenly the transformer size doubles. I still do not see a few more chokes and mains transformers as a major issue, compared to the cost of decent Output and Interstage transformers. You need to decide if cost is an object or not.
If you want a "cheap" amp build a JE-Labs/Walton or Reichert 300B Amp or one with a 6SL7 SRPP/Mu-Follower Driver (aka Welborne Labs Laurel and DIY HiFisupply Kit), Stereo chassis, Hammond Iron and good generic PSU Electrolytics with the output stage using "WE Connection" with a 22uF...33uF Cap from +B to the 300B Cathode and 100uF/160V Electrolytic to ground. Makes a cheap and effective amp.
Or if you take my advise you use the E810F or similar as Input/Driver with a DN2540 Depletion Fet based active anode load and a Gridchoke (plus coupling cap), you can try both the 6SN7 and 6SL7 stage and the E810F/6S45 et al and make up your own mind.
Sayonara
DVDHack said:
With 100/120Hz the hum is much more audible than 50Hz hum. for reference, 1.4mV would be 66db below 1W (on the 8 Ohm tap and it scales to the other Taps). So, if your speakers have 85db/W/m you get a steady state Hum with an absolute SPL of 19db at 1m Distance, probably inaudible at the listening posiition. If you use 103db/W/m sensitive Beauhorn Virtuoso Speakers (and they are 103db/1W @ 100Hz) the hum level would be 37db at 1m Distance, I personally would find such levels of hum certainly objectionable, I of course cannot speake for you.
My current Amp's are running on speakers that are around 96db/1W/1m @ 100Hz and have around 0.2...0.5mV 100Hz from the use of AC heaters (they are 45 Push-Pull). I find this objectionable only if listening very quiet late at night.
DVDHack said:
You still keep the grounds common. This may cause more problems than you think. Also, suddenly the transformer size doubles. I still do not see a few more chokes and mains transformers as a major issue, compared to the cost of decent Output and Interstage transformers. You need to decide if cost is an object or not.
If you want a "cheap" amp build a JE-Labs/Walton or Reichert 300B Amp or one with a 6SL7 SRPP/Mu-Follower Driver (aka Welborne Labs Laurel and DIY HiFisupply Kit), Stereo chassis, Hammond Iron and good generic PSU Electrolytics with the output stage using "WE Connection" with a 22uF...33uF Cap from +B to the 300B Cathode and 100uF/160V Electrolytic to ground. Makes a cheap and effective amp.
Or if you take my advise you use the E810F or similar as Input/Driver with a DN2540 Depletion Fet based active anode load and a Gridchoke (plus coupling cap), you can try both the 6SN7 and 6SL7 stage and the E810F/6S45 et al and make up your own mind.
Sayonara
Not worried about cost but don't want to spend money for the sake of spending it either. I guess I find myself questioning some of the wisdom.
We use a series of transformers would with a stack of copper wire and then insist that silver wire for a few inches makes the world of difference. I'm not saying it ain't so just maintaining a healthy scepticism on th true benefits.
...yes I know about the silver would TFRs (thinks thats a plot to drive up demand for silver), I'm not going for those either.
Regards
Ralf
We use a series of transformers would with a stack of copper wire and then insist that silver wire for a few inches makes the world of difference. I'm not saying it ain't so just maintaining a healthy scepticism on th true benefits.
...yes I know about the silver would TFRs (thinks thats a plot to drive up demand for silver), I'm not going for those either.
Regards
Ralf
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.