Audiomix said:
Good evening Hans, thanks for your input... Impressive work with the point to point. I do disagree re the input - to quote the designer himself....
"Usually the gain of the preamplifiers is around 15 db, and yours is even 27 db, so it’s not easy. If you want to decrease the gain of GM 20, you can utilize an attenuator in parallel by input of GM 20, working on resistances R2, R3 and R4, R5. Doing so the dynamics will be very low, or disappears at all, this is important for you to know."
Not quite so easy. I've spoken to some prominent amp designers (Jukka Toloken, Allen Wright etc) who have all commented that the Graaf GM20 is a good place to start but major improvements are to be had with some tinkering...
Thankfully hi-fi is a journey and at the moment I have the funds to commission work from a guru... it's just a case of working out which direction to go in...
Both of the amps have the copper bars, one retrofitted... They keep my Italian apartment very cosy in the winter but it's a nightmare in the summertime!
The input resistor network sets the sensitivity of the amplifier.
By changing the values you can set the sensitivity to eg 0.5V, 1.0V, 2.0V and so on. The sensitivity is the rms voltage required to drive the amplifier to full output.
By changing these resistors you will not change the dynamic performance. The dynamic performance of an amplifier is the ratio between full output and the noise level. If you change the sensitivity, you don't change the noise level nor the maximum output level.
There is nothing to do with the power consumption / heat dissipation except reducing the idling current (bias) in the output tupes. This may affect the HiFi performace.
What you should do is getting the heat out of and away from the amplifier. Lowering the overall temperature will extend the operating life time a lot.
By changing the values you can set the sensitivity to eg 0.5V, 1.0V, 2.0V and so on. The sensitivity is the rms voltage required to drive the amplifier to full output.
By changing these resistors you will not change the dynamic performance. The dynamic performance of an amplifier is the ratio between full output and the noise level. If you change the sensitivity, you don't change the noise level nor the maximum output level.
There is nothing to do with the power consumption / heat dissipation except reducing the idling current (bias) in the output tupes. This may affect the HiFi performace.
What you should do is getting the heat out of and away from the amplifier. Lowering the overall temperature will extend the operating life time a lot.
Taking the heat away:
Here is an effective way of reducing the heat. A wooden block with blowers as base plate for the amplifier. The blowers are absolutely quiet. It will reduce the chassis temperature of a GM20 by 20°C. Can be used on any amplifier running too hot. The one shown here is a cool running PP amplifier. Put it on for demonstartion only.
Here is an effective way of reducing the heat. A wooden block with blowers as base plate for the amplifier. The blowers are absolutely quiet. It will reduce the chassis temperature of a GM20 by 20°C. Can be used on any amplifier running too hot. The one shown here is a cool running PP amplifier. Put it on for demonstartion only.
dpaws said:Evening all - I'm afraid I'm a newbie...
I have a pair of Graafs GM20's (schematic via link) but the constant tweeking of bias and offset is driving me mad!
https://cid-a332bc3aefc89f6b.skydrive.live.com/self.aspx/Public/GM20 schematic.pdf
Is there a way of adding an auto bias circuit?
Also, can anyone suggest any tweeks / improvements? I've scoured the net but nothing forthcoming.
Ahh, can I reduce the gain / adapt the present circuit to suit a higher input voltage without using a pot and thus destroying the dynamics? I'm using a Weiss Medea DAC with +27dBu via XLR's....
Thanks guys
for several years I have this "GM 20" exemplare for repair. The pdf schematic I had prepared in circuitmaker (backengeneeríng), because there are don't available by Graaf.
To enhance reliability and lifetime a little, I reduce the idle current to the lowest of the mentioned values in my schematic. Additional I connect the main connection so to the primary windings, that I get the lowest possible value of anode voltage. But it is still to high.
according to datasheet max value must be arround 90V for the 6C33C - consequently the next step is to create a separate external power supply only for the output stage. Independent of this many addiional design mistakes there are, e. g. the very high dc impedance conditions arround the inputs of 6C33C. A friend of me will create and diy a GM20 clone for considerably more stable operation and better sonic quality
Concerning the quality of 6C33C - please note, there are two logos from Ulianov: an old and a newer (to see by currently production - please go to
http://www.ominous-valve.com/russtube.html
fourth logo. Some users tell me, by 6C33C devices from the old logo they get enhanced sonic quality and realibility/lifetime. What your experience in this case?
Hmm....
Thinking of your 'problem'
How old are the 6C33C-B, and what has changed since the amp was 'stable'...
(Or has it always been a 'Drifter'??)
IME, the 6C33 needs Very Careful burn-in, or it will drift--prolly through most/all its life....
I checked out the Graaf CCT above, and I believe an Auto-Bias is possible,--But will defo use sand as part/most of it!
--Although Should Not Be Necessary.....!
6C33C is sensitive to Heater-Volts, (with regards to Bias-Setting, just like any high perveyance type) and easily burns its sockets, particularly the high-current heater-pins....
In the Graaf above, the heat dissipation per Plate is 34.5W, and that is NOT including the Heater dissipation, in each bottle so those tubes get VERY hot, and even with perfect quality sockets the pin/socket connection can easily develop issues with time...
Slight high-resistance on a heater-pin and before you know it, the bias has altered, and if Cathode runs cooler than it should for any length of time due to poor heater-pin contact, will cause it to go low emission, and again,--Bias will alter...
In my use of these tubes in Circlotron I use series heaters, running off 25V supply, never had a high-resistance socket to pin issue...
--Also, I use a Very low bias of only 50mA per tube, so everything runs 'cool' --for a 6C33C Circlotron OTL and Ive not had to adjust bias in around 2 years!
--Hans who also is on this Forum has his own OTL design based on SRPP if I recall correctly, who also does not have Bias drift even after years of service--So it cant actually be the Tubes causing this bad rep, Rather the way they are being Used....
So,-
Check very carefully the heater pin connections of the 6C33C, and rectify any issues found. If the sockets are found to be burned or the spring-contacts have a low tension, Replace the socket completely.--Cleaning May help for a while, but its putting off the inevitable...
New tubes should be burned in in the following way--(Not much point in doing this to used tubes....)
Heaters only for 12-24 hours....
Cool down.
heaters on for 1/2 hour or so, Then draw plate current of 400 to 500mA or so--BUT KEEP Plate diss to a minimum....(Cant do this in the Amp, I made a set-up to do this with a 13V transformer, a bridge-rec, a couple of sockets and a 120W 240V bulb....)
Continue drawing 400-500mA for at least 3 hours. Usually I leave them for around 8 hours....
Cool down.--Tubes will then be 'burned-in' and be reasonably stable, to plus or minus 5mA or so when in use
Put in amp and set up bias. Re check/adjust Bias after a half-hour.
Be aware, the 6C33C bias will alter for 15 mins from cold switch on until it has thermally stabilised--This is normal.....
Thinking of your 'problem'
How old are the 6C33C-B, and what has changed since the amp was 'stable'...
(Or has it always been a 'Drifter'??)
IME, the 6C33 needs Very Careful burn-in, or it will drift--prolly through most/all its life....
I checked out the Graaf CCT above, and I believe an Auto-Bias is possible,--But will defo use sand as part/most of it!
--Although Should Not Be Necessary.....!
6C33C is sensitive to Heater-Volts, (with regards to Bias-Setting, just like any high perveyance type) and easily burns its sockets, particularly the high-current heater-pins....
In the Graaf above, the heat dissipation per Plate is 34.5W, and that is NOT including the Heater dissipation, in each bottle so those tubes get VERY hot, and even with perfect quality sockets the pin/socket connection can easily develop issues with time...
Slight high-resistance on a heater-pin and before you know it, the bias has altered, and if Cathode runs cooler than it should for any length of time due to poor heater-pin contact, will cause it to go low emission, and again,--Bias will alter...
In my use of these tubes in Circlotron I use series heaters, running off 25V supply, never had a high-resistance socket to pin issue...
--Also, I use a Very low bias of only 50mA per tube, so everything runs 'cool' --for a 6C33C Circlotron OTL and Ive not had to adjust bias in around 2 years!
--Hans who also is on this Forum has his own OTL design based on SRPP if I recall correctly, who also does not have Bias drift even after years of service--So it cant actually be the Tubes causing this bad rep, Rather the way they are being Used....
So,-
Check very carefully the heater pin connections of the 6C33C, and rectify any issues found. If the sockets are found to be burned or the spring-contacts have a low tension, Replace the socket completely.--Cleaning May help for a while, but its putting off the inevitable...
New tubes should be burned in in the following way--(Not much point in doing this to used tubes....)
Heaters only for 12-24 hours....
Cool down.
heaters on for 1/2 hour or so, Then draw plate current of 400 to 500mA or so--BUT KEEP Plate diss to a minimum....(Cant do this in the Amp, I made a set-up to do this with a 13V transformer, a bridge-rec, a couple of sockets and a 120W 240V bulb....)
Continue drawing 400-500mA for at least 3 hours. Usually I leave them for around 8 hours....
Cool down.--Tubes will then be 'burned-in' and be reasonably stable, to plus or minus 5mA or so when in use
Put in amp and set up bias. Re check/adjust Bias after a half-hour.
Be aware, the 6C33C bias will alter for 15 mins from cold switch on until it has thermally stabilised--This is normal.....
Aww many thanks for all of your thoughts - let me chew it all over and I'll return in a couple of days.
My "problem" has arisen over my intended application with an active (or passive line level) crossover before the amps - hence I need them to be relaible stable...
They've been retired for the summer as I'm living in Italy as it gets crazy hot with a pair of GM20's when it's 32'C outside... I do feel a rebuild coming on!
Ci vediamo! Steve
My "problem" has arisen over my intended application with an active (or passive line level) crossover before the amps - hence I need them to be relaible stable...
They've been retired for the summer as I'm living in Italy as it gets crazy hot with a pair of GM20's when it's 32'C outside... I do feel a rebuild coming on!
Ci vediamo! Steve
for several years I have this "GM 20" example for repair. The pdf schematic I had prepared in circuitmaker (backengeneeríng), because there are don't available by Graaf.
.... A friend of me will create and diy a GM20 clone for considerably more stable operation and better sonic quality....
Please is there any news of the redesign??
It's winter time and so the apartment needs some heating
for several years I have this "GM 20" example for repair. The pdf schematic I had prepared in circuitmaker (backengeneeríng), because there are don't available by Graaf.
.... A friend of me will create and diy a GM20 clone for considerably more stable operation and better sonic quality....
Please is there any news of the redesign??
It's winter time and so the apartment needs some heating
As I know, the diyaudio member "Jon Lord" currently perform a redesign. You find Jon Lord by #145 about
http://www.diyaudio.com/forums/solid-state/114865-densen-amp-15.html
An other way it could be to create a circlotron buffer by using of MOSFETs instead of expensive 6C33C triodes at the output - simulation results of me you can find by post #17 about
http://www.diyaudio.com/forums/soli...-descr-arround-q12a-wanted-2.html#post1997611
my unity gain circlotron prototype runs currently by 2x 10V voltage supply and 1A idle current. The sonic transmission is very fine.
Last edited:
Are there any news in this matter?.... A friend of me will create and diy a GM20 clone for considerably more stable operation and better sonic quality...
Have you contact the member "Jon Lord" ?
Last edited:
Comments from Allen Wright
Hello dpaws,
I know this thread is already from 2009 so quite some time ago and I also know that Allen Wright passed away in 2011 unfortunately.
Nevertheless I would be keen to hear if you got concrete recommendations from Allen then how to improve the Graaf GM20. I do have his RTP3 preamp and I also read his tube preamp cookbook.
From this reading I would assume that he advised to put CCS in the cathodes of the 6922/EF184 rather than using resistors and probably apply a super reg ( ccs in combination with a shunt reg ) voltage regulator at least in the driver stages of the EF184 tubes. This cannot be done with the 6922 differential input amp as they are fed by the circlotron HT supply.
By the way I also use a clown of the Graaf GM20 build by a tube DIY'er with more than 30 year experience offering them from time to time on german Ebay. They are however improved and much more stable compared to the Graafs and build completely hardwired as mono blocks. Apart from that Bias and DC offset can continuously be monitored by build-in analog meters which is a big advantage compared to the Graaf GM20.
See here from the last auction:
OTL Rohrenverstarker | eBay
Hope to hear from you.
Thanks
airtangent
Hello dpaws,
I know this thread is already from 2009 so quite some time ago and I also know that Allen Wright passed away in 2011 unfortunately.
Nevertheless I would be keen to hear if you got concrete recommendations from Allen then how to improve the Graaf GM20. I do have his RTP3 preamp and I also read his tube preamp cookbook.
From this reading I would assume that he advised to put CCS in the cathodes of the 6922/EF184 rather than using resistors and probably apply a super reg ( ccs in combination with a shunt reg ) voltage regulator at least in the driver stages of the EF184 tubes. This cannot be done with the 6922 differential input amp as they are fed by the circlotron HT supply.
By the way I also use a clown of the Graaf GM20 build by a tube DIY'er with more than 30 year experience offering them from time to time on german Ebay. They are however improved and much more stable compared to the Graafs and build completely hardwired as mono blocks. Apart from that Bias and DC offset can continuously be monitored by build-in analog meters which is a big advantage compared to the Graaf GM20.
See here from the last auction:
OTL Rohrenverstarker | eBay
Hope to hear from you.
Thanks
airtangent
GM20 DIY
Hello Sir,
I buit two OTL amplifiers be Tim Millow and Japanese guy, now I am going to make GM20 clone. I have a circuit diagram of it, want to know if the scheme is correct so there are questions for me in some places. I want to compare the circuit diagrams, maybe you have a GM20 schematics too? I can send my schematics to you. kest2010@gmail.com.
Kestutis
Hello Sir,
I buit two OTL amplifiers be Tim Millow and Japanese guy, now I am going to make GM20 clone. I have a circuit diagram of it, want to know if the scheme is correct so there are questions for me in some places. I want to compare the circuit diagrams, maybe you have a GM20 schematics too? I can send my schematics to you. kest2010@gmail.com.
Kestutis
Last edited by a moderator:
Hello Kestutis,
Great to get finally a response on my Graaf GM20 chat!
Yes I do have a detailed GM20 schematic and the best one I could find so far can be found here on diyaudio as well:
Graaf OTL 20 amp
See on top of this page the attached pdf file from member Nightpuma.
Thank you for you offer but I do have already the Tim Mellow schematic and personally I would prefer the Graaf design which can be slightly improved as I said already above.
Of course the ebay link doesn't work anymore but I found a site showing them from
outside and inside, not mine but very similar giving a good impression how they are made:
Graaf GM20 OTL Clone kaufen auf Ricardo
Regards
airtangent
Great to get finally a response on my Graaf GM20 chat!
Yes I do have a detailed GM20 schematic and the best one I could find so far can be found here on diyaudio as well:
Graaf OTL 20 amp
See on top of this page the attached pdf file from member Nightpuma.
Thank you for you offer but I do have already the Tim Mellow schematic and personally I would prefer the Graaf design which can be slightly improved as I said already above.
Of course the ebay link doesn't work anymore but I found a site showing them from
outside and inside, not mine but very similar giving a good impression how they are made:
Graaf GM20 OTL Clone kaufen auf Ricardo
Regards
airtangent
Hello satsek18,
Improvements are CRC power supply with 4 * 1800uF, so in total 7200uF
( Graaf GM20 has just one big 4700 uF cap ) capacity per side. Two pots in the bias circuit, one for adjusting the negative grid voltage for bias and the other connected to the output of the first pot for balancing the bias over both 6C33 tubes. A better way in order to minimize residual DC offset in my opinion. All adjustments can be done from outside and by means of the analog meters bias and DC offset can be monitored continously.
Regards,
airtangent
Improvements are CRC power supply with 4 * 1800uF, so in total 7200uF
( Graaf GM20 has just one big 4700 uF cap ) capacity per side. Two pots in the bias circuit, one for adjusting the negative grid voltage for bias and the other connected to the output of the first pot for balancing the bias over both 6C33 tubes. A better way in order to minimize residual DC offset in my opinion. All adjustments can be done from outside and by means of the analog meters bias and DC offset can be monitored continously.
Regards,
airtangent
GM20 improvements
Regards,
Kestas
Thanks a lot for an ansver. What about MV1 NTC, is it needed at all? What is value ?
Regards,
Kestas
I made my own GM20.
It's better, more power and stable without running warm.
Hence the name Am30.
Made by Audiomix (Am) and more than 30 Watts output.
It's better, more power and stable without running warm.
Hence the name Am30.
Made by Audiomix (Am) and more than 30 Watts output.
An externally hosted image should be here but it was not working when we last tested it.
Kay Pirinha,
The link does work but you must use open link in a new tab or new window and then you can see it.
Audiomix,
Looks great your AM20. I would be very curious to hear from you how you get 30 Watt out of it.
As far as I know the GM20 uses a bias of 200 – max 250 mA at around 160V anode voltage.
So I suppose that you make use of another bias setting with somewhat higher anode dissipation.
And then due to your active continuous cooling by means of the fans you probably will not sacrifice all too much w.r.t. life span of the 6c33 tubes ( usually 2000 – 2500 hrs ) I guess, right?
When you say it's better what else did you improve?
Kind regards,
airtangent
The link does work but you must use open link in a new tab or new window and then you can see it.
Audiomix,
Looks great your AM20. I would be very curious to hear from you how you get 30 Watt out of it.
As far as I know the GM20 uses a bias of 200 – max 250 mA at around 160V anode voltage.
So I suppose that you make use of another bias setting with somewhat higher anode dissipation.
And then due to your active continuous cooling by means of the fans you probably will not sacrifice all too much w.r.t. life span of the 6c33 tubes ( usually 2000 – 2500 hrs ) I guess, right?
When you say it's better what else did you improve?
Kind regards,
airtangent