Mimesis Build
Hi Eivind,
Yesterday I have (almost) completed one channel of the Mimesis 3 Build.
The transformer is rated as 160VA 40V AC.
I measured the Secondary AC voltage whilst playing music at normal listening level, it shows 43.5V AC.
Voltage for output MOSFET is 57.5V DC
Voltage for the frontend (Input & VAS) are +59.3V & -60.7V
For T1, I used a pair of 2SK170BL
T3 & T4: Fairchild BC546B
T5 & T6: Fairchild BC182B
T11: On-semi MPSA42
T12: On-semi MPSA92
T15 & T16: Reneses 2SK1058 (TO3p)
T17 & T18: Reneses 2SJ162 (TO3p)
The rest are just Fairchild MPSA42/92
The T7, T8, T9 & T10 are running hot. I super-glued a small PCB to T11 & T12 so they are running cool.
I also experienced the same problem of drifting DC offset, it varies by as much 40 to 50mV. I used 10R as R36 & it made no difference.
I think that problem is with the front-end power supply and the voltage fluctuates a fair bit (by a couple of Volts) & that would affect the DC offset.
Listening impression:
This is my third L-MOSFET amplifier & I prefer the sound of L-MOSFET to BJT output devices.
(1) I built Dr Bora's Sigma Amp: http://bas.elitesecurity.org/SIGMA-Schematic.pdf
The Sigma amp sounds smooth & musical. Bass is good but the high lacks a bit of details.
(2) I built a amp with 2 pairs of L-MOSFET & National Semi's LMe49830 driver chip.
The LMe49830 sounds a bit more detailed than the Sigma amp & it sounds a bit sterile.
(3) Mimesis 3: I am only doing a bench test with one pair of L-MOSFET with one board at the moment. It sounds smooth & detailed, though the bass lacks some punch.
Alexmm has done a great job with the PCB layout, the spacing of components are just right & pads are generously sized.
Attached are some pictures:
Cheers, Stanley
Hi Eivind,
Yesterday I have (almost) completed one channel of the Mimesis 3 Build.
The transformer is rated as 160VA 40V AC.
I measured the Secondary AC voltage whilst playing music at normal listening level, it shows 43.5V AC.
Voltage for output MOSFET is 57.5V DC
Voltage for the frontend (Input & VAS) are +59.3V & -60.7V
For T1, I used a pair of 2SK170BL
T3 & T4: Fairchild BC546B
T5 & T6: Fairchild BC182B
T11: On-semi MPSA42
T12: On-semi MPSA92
T15 & T16: Reneses 2SK1058 (TO3p)
T17 & T18: Reneses 2SJ162 (TO3p)
The rest are just Fairchild MPSA42/92
The T7, T8, T9 & T10 are running hot. I super-glued a small PCB to T11 & T12 so they are running cool.
I also experienced the same problem of drifting DC offset, it varies by as much 40 to 50mV. I used 10R as R36 & it made no difference.
I think that problem is with the front-end power supply and the voltage fluctuates a fair bit (by a couple of Volts) & that would affect the DC offset.
Listening impression:
This is my third L-MOSFET amplifier & I prefer the sound of L-MOSFET to BJT output devices.
(1) I built Dr Bora's Sigma Amp: http://bas.elitesecurity.org/SIGMA-Schematic.pdf
The Sigma amp sounds smooth & musical. Bass is good but the high lacks a bit of details.
(2) I built a amp with 2 pairs of L-MOSFET & National Semi's LMe49830 driver chip.
The LMe49830 sounds a bit more detailed than the Sigma amp & it sounds a bit sterile.
(3) Mimesis 3: I am only doing a bench test with one pair of L-MOSFET with one board at the moment. It sounds smooth & detailed, though the bass lacks some punch.
Alexmm has done a great job with the PCB layout, the spacing of components are just right & pads are generously sized.
Attached are some pictures:
Cheers, Stanley
Attachments
Hi my friend Liliya
Not confused as a first step is easier and quicker to investigate problems rather than to take off parts from the top side now that verified all the voltages I have reinstall the parts on the upper side.
I had a doubt about how to affect the symmetry of the doubler voltage on dc offset
Now I draw my conclusion (no problem at all)
Positive rail 84.05V
Negative rail 82.03V
The negative side has a greater load at 1mA - 1,2 mA during ZD1-T6-R12
Also i use 10 pf Kemet exactly on C3 - C5
without pull out any of the capacitor c3,c5,c6,c7 (c3-c5 =14,7 pf).
Not confused as a first step is easier and quicker to investigate problems rather than to take off parts from the top side now that verified all the voltages I have reinstall the parts on the upper side.
I had a doubt about how to affect the symmetry of the doubler voltage on dc offset
Now I draw my conclusion (no problem at all)
Positive rail 84.05V
Negative rail 82.03V
The negative side has a greater load at 1mA - 1,2 mA during ZD1-T6-R12
Also i use 10 pf Kemet exactly on C3 - C5
without pull out any of the capacitor c3,c5,c6,c7 (c3-c5 =14,7 pf).
GM amp
I think that the HMA 7500 MKII is a completly different amplifier.
What makes the GM amp interesting is:
All DC design. No capacitor in the feedback loop. High sleve rate and fast rise time. 0 degree phase error. Less than 200ns flat group delay. Very flat frequency respons. Very low noise figure.
Most other designs do not have this as Jacco Vermeulen pointed out earlier.
I think that the HMA 7500 MKII is a completly different amplifier.
What makes the GM amp interesting is:
All DC design. No capacitor in the feedback loop. High sleve rate and fast rise time. 0 degree phase error. Less than 200ns flat group delay. Very flat frequency respons. Very low noise figure.
Most other designs do not have this as Jacco Vermeulen pointed out earlier.
Sng001
((I think that problem is with the front-end power supply and the voltage fluctuates a fair bit (by a couple of Volts) & that would affect the DC offset))
I do not think this is the problem.
if you have matched transistors 2SK170BL and T3,T4 and you have full symmetry currents on R16, R17, R18. You have no problem.
R16 =1,960V --- 13,06 ma
R17= 2,100 ---6,36 ma
R18=2,200----6,6 ma
((I think that problem is with the front-end power supply and the voltage fluctuates a fair bit (by a couple of Volts) & that would affect the DC offset))
I do not think this is the problem.
if you have matched transistors 2SK170BL and T3,T4 and you have full symmetry currents on R16, R17, R18. You have no problem.
R16 =1,960V --- 13,06 ma
R17= 2,100 ---6,36 ma
R18=2,200----6,6 ma
I did the voltage measurement:
R16= 1.969V --- 13.13 ma
R17= 2.132V --- 6.46 ma
R18= 2.200V --- 6.67 ma
The front-end power supply was measured to be +60.0 / -58.7 in the morning and it just measured to be +58.0 / -56.9 so it does fluctuate by 2V.
BTW, I have tried my best to match the Idss of the 2SK170BL and Hfe of T3 & T4.
The DC offset varies from -15mV to +24mV. To me it is still acceptable, it is just different from the other amp that I build.
I will build a better front power supply as the drop resistors (R29 & R34) are running too hot for long term reliability. BTW, AlexMM thought ahead & put some vent holes on the PCB.
R16= 1.969V --- 13.13 ma
R17= 2.132V --- 6.46 ma
R18= 2.200V --- 6.67 ma
The front-end power supply was measured to be +60.0 / -58.7 in the morning and it just measured to be +58.0 / -56.9 so it does fluctuate by 2V.
BTW, I have tried my best to match the Idss of the 2SK170BL and Hfe of T3 & T4.
The DC offset varies from -15mV to +24mV. To me it is still acceptable, it is just different from the other amp that I build.
I will build a better front power supply as the drop resistors (R29 & R34) are running too hot for long term reliability. BTW, AlexMM thought ahead & put some vent holes on the PCB.
Jeffrey,
the current through the front end remains identical, regardless of the rail voltage level.
Means you have to adjust the value of the (4W?) power resistor before the BD249/250 series devices, if fed by 45Vac instead of 60Vac.
The output stage bias level is a major shortcoming of the Goldmund amp, imo.
GM wastes power in the regulator sections (which actually are not even proper voltage regulators), but goes total Eco in the output stage, squeamish on heatsinks for a powerfull amp.
To my ears, each low quiescent level MOSFET output stage sounded very harsh.
Fast, accurate to the extreme, but harsh sounding amps in the voice and up freq. range, is typical German school of the '80s and '90s. (as the lingo can be, if desired. you'd think Mr Yvon was Elzas born)
Bare minimum bias level for lateral MOSFETs is 100mA each, but >1.2V across the gates for 125-150mA is more preferable.
For an explanation, look at the 75C curve in the Vgs/Id graph of both N- and P-channel device.
Last edited:
thank you ... thank you very much
Some comments on a stable and low DC on the output witch I also have not managed to cure:.
As sng001, I tried to replace T3/T4 with matched BC 546B.No change.
I had hoped that by replacing the power supply to the front end with a separate and variable power unit,would remove the DC problem.
The effect of having the exact same voltage on each side of the front end (+-80v DC), was not as pronounced as I had hoped for. But of course a separate power have other advantages over the original voltage doubler.
I might also mention that T1 is 2N5564 (not 65) with me, since the use of 2SK170 BL have being mentioned as a possible explanation for the DC problem.
My conclusion is that the problem lies elsewhere, and I've not been able to locate it.
Someone on the forum here have pointed out that the low value of R25 (0.22 Ohm/ "short-circuiting" the coil) can influence on DC. I intend to try with 2.2 Ohm for R25 later on and see what happens.
sng001: Your listening tests indicate the same as what I have experienced (highly subjective): Goldmund play highly detailed, open and comfortable (not harsh) in the range from 600 Hz and upwards in my setup with electronic crossover. I have also listened to the "middle bass" on a full range speaker and find it very potent and satisfying. I have not had the opportunity to test out the lowest bass the same way.
The bias in my Goldmund is very close to what jacco recommends.
Eivind Stillingen
As sng001, I tried to replace T3/T4 with matched BC 546B.No change.
I had hoped that by replacing the power supply to the front end with a separate and variable power unit,would remove the DC problem.
The effect of having the exact same voltage on each side of the front end (+-80v DC), was not as pronounced as I had hoped for. But of course a separate power have other advantages over the original voltage doubler.
I might also mention that T1 is 2N5564 (not 65) with me, since the use of 2SK170 BL have being mentioned as a possible explanation for the DC problem.
My conclusion is that the problem lies elsewhere, and I've not been able to locate it.
Someone on the forum here have pointed out that the low value of R25 (0.22 Ohm/ "short-circuiting" the coil) can influence on DC. I intend to try with 2.2 Ohm for R25 later on and see what happens.
sng001: Your listening tests indicate the same as what I have experienced (highly subjective): Goldmund play highly detailed, open and comfortable (not harsh) in the range from 600 Hz and upwards in my setup with electronic crossover. I have also listened to the "middle bass" on a full range speaker and find it very potent and satisfying. I have not had the opportunity to test out the lowest bass the same way.
The bias in my Goldmund is very close to what jacco recommends.
Eivind Stillingen
Last edited:
Nikosokey
ECF10N16-ECF10P16 OR BUZ9000-905 ?
I have no personal experience with the output transistors, which is the optimum. I'm not even sure if I would be able to distinguish between the two products you mention.
Although I chosed to go for ECF. The choice was taken from the statements of several on this forum that claimed to be able to "jugde" the sonically best of the two. I have seen this discussed both in this thread and in other threads that many would prefer ECF.
What have you placed on top of the transformer?
Protection are in very quick after I activate the main power, less than a second (I have not measured)
Have you noticed what Alex (or was it Bigpanda) said about T4??
Eivind Stillingen
It will be exciting to watch your continuation of your Goldmund clone.
ECF10N16-ECF10P16 OR BUZ9000-905 ?
I have no personal experience with the output transistors, which is the optimum. I'm not even sure if I would be able to distinguish between the two products you mention.
Although I chosed to go for ECF. The choice was taken from the statements of several on this forum that claimed to be able to "jugde" the sonically best of the two. I have seen this discussed both in this thread and in other threads that many would prefer ECF.
What have you placed on top of the transformer?
Protection are in very quick after I activate the main power, less than a second (I have not measured)
Have you noticed what Alex (or was it Bigpanda) said about T4??
Eivind Stillingen
It will be exciting to watch your continuation of your Goldmund clone.
Thank you but you first took the initiative to build the amplifier and show us how it works. Personal and I had and have doubts for the proper functioning of the amplifier.Gave me a big boost in project amplifier at some Post the official datasheet of GOLMUND .Of course I will mention the Post gives us, Mr JAcco Vermeulen great help.
28-year old Mr Nagys, with more than 10 years experience in (selling) High-End audio, may disagree.
Plenty of very experienced guys have posted on this thread, who are worth the read, such as ;
-Operating System Tripper, able enough to fix very complex exotic amps.
-Jam, who had his eyes on close to every amp schematic.
-Mr Wahab
-Oompie Nico Ras
If you're going for +/-80V rails, first priority is to choose output devices with 180Vgs minimum, the equivalent of J55/K175.
Neither Magnatec or Exicon manufacture 180V types (no such thing as an EC-10N18 or a BUZ905.5
), so you'd have to go for EC-10N20/10P20, or BUZ901/906.
BUZ905/EC-10N16 are good for 70V rails.
Plenty of very experienced guys have posted on this thread, who are worth the read, such as ;
-Operating System Tripper, able enough to fix very complex exotic amps.
-Jam, who had his eyes on close to every amp schematic.
-Mr Wahab
-Oompie Nico Ras
If you're going for +/-80V rails, first priority is to choose output devices with 180Vgs minimum, the equivalent of J55/K175.
Neither Magnatec or Exicon manufacture 180V types (no such thing as an EC-10N18 or a BUZ905.5
), so you'd have to go for EC-10N20/10P20, or BUZ901/906.BUZ905/EC-10N16 are good for 70V rails.
still at it , ehhh !!
Right,Anal matching of the cascode or input pairs will make no difference. Amp (below 1) will stay at +/- 1mV with 2 "mismatched" input transistors , ambient to hairdryer temperature swings , or mismatched supply rails. Matching is more important for final sound quality , so besides testing (playing).. match your IPS.
I know the goldmund is very fast , but if that speed compromises "basics" like the rock solid stability of the DC offset .... unacceptable. I tried my EX , which is just like the goldmund without a FET input. Same 1mV offset , even tried unbalanced rails , no more than 1-2mv offset change. I HAVE seen "jumpy" offsets and Unstable Vbe readings with one of my modules at or near oscillation. I did this purposely just to see what "errata" I might experience.
I would "slow the amp down" (use temporary 68-100pf Cdom's) just to see whether the "errata" disappears.
OS
Right,Anal matching of the cascode or input pairs will make no difference. Amp (below 1) will stay at +/- 1mV with 2 "mismatched" input transistors , ambient to hairdryer temperature swings , or mismatched supply rails. Matching is more important for final sound quality , so besides testing (playing).. match your IPS.
That is after NFB takeoff , it would have no effect on offset.
I know the goldmund is very fast , but if that speed compromises "basics" like the rock solid stability of the DC offset .... unacceptable. I tried my EX , which is just like the goldmund without a FET input. Same 1mV offset , even tried unbalanced rails , no more than 1-2mv offset change. I HAVE seen "jumpy" offsets and Unstable Vbe readings with one of my modules at or near oscillation. I did this purposely just to see what "errata" I might experience.
I would "slow the amp down" (use temporary 68-100pf Cdom's) just to see whether the "errata" disappears.
OS
