Nisbeth said:
Most of the higher power flat-pack transistors are relatively modern devices with a fairly high ft. This can cause problems with instability when used in the JLH unless additional compensation is used which would negate one of the benefits of this amp, namely an open-loop bandwidth that extends across most, if not all, of the audio spectrum.
The potential for instability depends a lot on the version being built and the physical layout. I have heard from one constructor who has successfully used the 2SC5200 in a 1969 version (which seems to be slightly more tolerant of higher ft devices, possibly because the bootstrap becomes less effective at high frequencies causing Q3 to have a higher collector load). However, other constructors have reported oscillation when using higher ft transistors, particularly with later versions of the amp.
The only low ft (4MHz) flat-pack transistor I have found (though I haven't checked all of the Sanken portfolio) is On-Semi's MJL21194, though unfortunately this device does not have as high a gain as I would like (but it is more linear than the MJ15003).
Why not TO3 ?
Hi Nisbeth,
why not use transistors in TO 3 pak for the JLH ?
Ich have checked many transistors in a JLH Schematic in Spice and the MJ11016 brings the best THD and IM Distortions.
All other Transistors (MJ15003, MJL21194, MJL3281A and MJL4281A) have a worse THD factor of 1:2 and more.
The Sanken Transistors i couldn´t test. because i can´t find spice models of them.
Somebody knows where i can find Sanken Spice models ?
cu
small_boy
Hi Nisbeth,
why not use transistors in TO 3 pak for the JLH ?
Ich have checked many transistors in a JLH Schematic in Spice and the MJ11016 brings the best THD and IM Distortions.
All other Transistors (MJ15003, MJL21194, MJL3281A and MJL4281A) have a worse THD factor of 1:2 and more.
The Sanken Transistors i couldn´t test. because i can´t find spice models of them.
Somebody knows where i can find Sanken Spice models ?
cu
small_boy
> I have checked many transistors in a JLH Schematic in Spice and the MJ11016 brings the best THD and IM Distortions.
I wonder if your Spice model simulates hfe changes with current.
If one is keen on high hfe (which at least in theory would reduce distortion), may I suggest connecting a Sziklai Darlington (see Horowitz & Hill, Fig. 2.62) using MJ21194 (or even MJL21194 plastic packages as Geoff suggested) and 2SA1358. One could easily get a gain of 7500 at operating temperature, and would make the lower gain of the MJL version a non issue. Linearity would be miles better than MJ11016. And I am telling from my own measurements.
Patrick
I wonder if your Spice model simulates hfe changes with current.
If one is keen on high hfe (which at least in theory would reduce distortion), may I suggest connecting a Sziklai Darlington (see Horowitz & Hill, Fig. 2.62) using MJ21194 (or even MJL21194 plastic packages as Geoff suggested) and 2SA1358. One could easily get a gain of 7500 at operating temperature, and would make the lower gain of the MJL version a non issue. Linearity would be miles better than MJ11016. And I am telling from my own measurements.
Patrick
@Nisbeth :
Ok, thats right. Flatpacks are better integrated in a PCB layout.
@EUVL
I use a small modified version of the JLH design that can driven speaker with impedance smaller then 4 Ohm. Because my intent is to driven later very complex speaker (Electrostatical Speaker with lower Impedance of 2 ohm in High tone). For this the MJ11016 are better (look at measurements)
Another problem is the price of the MJL21194 or MJ21194 in germany. I see that you come also from Germany. I have look where i can get the MJL21194 and i found only by Segor Electronics the MJL21194, but the price of over 6 Euro is very high. By Reichelt i can get the MJ11016 for only 2.20 Euro.
Can you tell me where i can geht the MJ21194 or MJL21194 for a low price in Germany ( i need for my design over 40 of these Transistors!) ?
Here measurements of the Modified JLH with MJ11016 and MJ21194. Every Transistor has 0,9 A bias current.
Modified JLH with three pairs of MJ11016
Watt Dist. at 8 Ohm Dist at 4 Ohm Dist at 2 Ohm
---------------------------------------------------------------------------------------
50 mW 0,001% 0,000% 0,001%
1 W 0,004% 0,002% 0,006%
5 W 0,009% 0,005% 0,015%
10 W 0,013% 0,007% 0,025%
50 W 0,041% 0,033% 0,275%
max power :
at 8 Ohm - 0.170% Dist. - 83 Watt (can more but its cut off at +/- 34 Volt PSU)
at 4 Ohm - 0.443% Dist - 130 Watt (can more but its cut off at +/- 34 Volt PSU)
at 2 Ohm - 1% Dist - 153 Watt
Modified JLH with three pairs of MJ21194
Watt Dist. at 8 Ohm Dist at 4 Ohm Dist at 2 Ohm
---------------------------------------------------------------------------------------
50 mW 0,009% 0,005% 0,005%
1 W 0,041% 0,025% 0,030%
5 W 0,094% 0,057% 0,078%
10 W 0,138% 0,082% 0,139%
50 W 0,413% 0,274% ***
*** Cut off in Spice. MJ21194 cant driven ist.
Another way is to bridge two modified versions of the first Schematic with MJ11016 ( I know, it is overkilled ;-))
Watt Dist. at 8 Ohm Dist at 4 Ohm Dist at 2 Ohm
---------------------------------------------------------------------------------------
1 W 0,002% 0,003% 0,012%
5 W 0,005% 0,007% 0,030%
20 W 0,010% 0,019% 0,078%
50 W 0,018% 0,050% 0,352%
100 W 0,038% 0,223% 0,803%
Max Power at 8 Ohm 232 W with 0,650% Dist. (can more but its cut off at +/- 34 Volt PSU)
Max Power at 4 Ohm 315 W with 0,940% Dist. (can more but its cut off at +/- 34 Volt PSU)
Max Power at 2 Ohm 144 W with 1,135% Dist.
I know that a Spice simulation is one thing, the reality another. ;-))
Greeting from Germany
small_boy
Ok, thats right. Flatpacks are better integrated in a PCB layout.
@EUVL
I use a small modified version of the JLH design that can driven speaker with impedance smaller then 4 Ohm. Because my intent is to driven later very complex speaker (Electrostatical Speaker with lower Impedance of 2 ohm in High tone). For this the MJ11016 are better (look at measurements)
Another problem is the price of the MJL21194 or MJ21194 in germany. I see that you come also from Germany. I have look where i can get the MJL21194 and i found only by Segor Electronics the MJL21194, but the price of over 6 Euro is very high. By Reichelt i can get the MJ11016 for only 2.20 Euro.
Can you tell me where i can geht the MJ21194 or MJL21194 for a low price in Germany ( i need for my design over 40 of these Transistors!) ?
Here measurements of the Modified JLH with MJ11016 and MJ21194. Every Transistor has 0,9 A bias current.
Modified JLH with three pairs of MJ11016
Watt Dist. at 8 Ohm Dist at 4 Ohm Dist at 2 Ohm
---------------------------------------------------------------------------------------
50 mW 0,001% 0,000% 0,001%
1 W 0,004% 0,002% 0,006%
5 W 0,009% 0,005% 0,015%
10 W 0,013% 0,007% 0,025%
50 W 0,041% 0,033% 0,275%
max power :
at 8 Ohm - 0.170% Dist. - 83 Watt (can more but its cut off at +/- 34 Volt PSU)
at 4 Ohm - 0.443% Dist - 130 Watt (can more but its cut off at +/- 34 Volt PSU)
at 2 Ohm - 1% Dist - 153 Watt
Modified JLH with three pairs of MJ21194
Watt Dist. at 8 Ohm Dist at 4 Ohm Dist at 2 Ohm
---------------------------------------------------------------------------------------
50 mW 0,009% 0,005% 0,005%
1 W 0,041% 0,025% 0,030%
5 W 0,094% 0,057% 0,078%
10 W 0,138% 0,082% 0,139%
50 W 0,413% 0,274% ***
*** Cut off in Spice. MJ21194 cant driven ist.
Another way is to bridge two modified versions of the first Schematic with MJ11016 ( I know, it is overkilled ;-))
Watt Dist. at 8 Ohm Dist at 4 Ohm Dist at 2 Ohm
---------------------------------------------------------------------------------------
1 W 0,002% 0,003% 0,012%
5 W 0,005% 0,007% 0,030%
20 W 0,010% 0,019% 0,078%
50 W 0,018% 0,050% 0,352%
100 W 0,038% 0,223% 0,803%
Max Power at 8 Ohm 232 W with 0,650% Dist. (can more but its cut off at +/- 34 Volt PSU)
Max Power at 4 Ohm 315 W with 0,940% Dist. (can more but its cut off at +/- 34 Volt PSU)
Max Power at 2 Ohm 144 W with 1,135% Dist.
I know that a Spice simulation is one thing, the reality another. ;-))
Greeting from Germany
small_boy
Where to get MJ21194 in Germany
Small Boy,
I have no doubt that you'll get lower distortion from simulation with MJ11016, because of the much higher hfe than MJ21194. I look for open loop linearity first, which is a different story.
If you just want to experiment (e.g. my idea with a self made Darlington), you could get 25 samples for USD15- including postage directly from On Semi (www.onsemi.com).
If you decide that you need more, then I suggest ordering from Digikey (www.digikey.com). They have link to their German website where prices are quoted in Euro including import duties. For 10 pcs of MJ21194, they charge Euro 2.84, for 100 pcs Euro 2.11. VAT (MwSt) on top, and they do charge something like 15 Euros handling charge, so it is not worth ordering from them if you do not have a large order to place. I like their service and their prices very much. Just check out prices for Panasonic FC caps and compare with Farnell !!!!
I don't know whether Fairchild have them. They also have a direct purchase service which I use a lot.
Let us know how you get on with the comparison, experimentally.
Schönen Gruss,
Patrick
Small Boy,
I have no doubt that you'll get lower distortion from simulation with MJ11016, because of the much higher hfe than MJ21194. I look for open loop linearity first, which is a different story.
If you just want to experiment (e.g. my idea with a self made Darlington), you could get 25 samples for USD15- including postage directly from On Semi (www.onsemi.com).
If you decide that you need more, then I suggest ordering from Digikey (www.digikey.com). They have link to their German website where prices are quoted in Euro including import duties. For 10 pcs of MJ21194, they charge Euro 2.84, for 100 pcs Euro 2.11. VAT (MwSt) on top, and they do charge something like 15 Euros handling charge, so it is not worth ordering from them if you do not have a large order to place. I like their service and their prices very much. Just check out prices for Panasonic FC caps and compare with Farnell !!!!
I don't know whether Fairchild have them. They also have a direct purchase service which I use a lot.
Let us know how you get on with the comparison, experimentally.
Schönen Gruss,
Patrick
JLH vs 'Class-T'
Assuming there might be another JLH owner out there (apart from me) who might be curious to know how a digital amplifier compares with the JLH class A, then this is for you.
This comparison is between a JLH using 'selected' quality components and a Sonic Impact 't-amp' (Tripath TA2024 chip), its volume pot/input/output capacitors replaced with Noble/M'cap 'audiophile'/Hovland 'musicap'.
My impressions...
Both sound smooth though the JLH is the smoother of the two.
The t-amp has better tonal accuracy and focus, the position of instruments more precise. The JLH sounding diffused in comparison, rather more relaxed, a little hazy and less detailed (despite its having better signal capacitors and a stepped attenuator).
Side by side, the t-amp sounds superior in many ways but on prolonged listening it is the JLH which makes the better music....somehow!
Tim.
Assuming there might be another JLH owner out there (apart from me) who might be curious to know how a digital amplifier compares with the JLH class A, then this is for you.
This comparison is between a JLH using 'selected' quality components and a Sonic Impact 't-amp' (Tripath TA2024 chip), its volume pot/input/output capacitors replaced with Noble/M'cap 'audiophile'/Hovland 'musicap'.
My impressions...
Both sound smooth though the JLH is the smoother of the two.
The t-amp has better tonal accuracy and focus, the position of instruments more precise. The JLH sounding diffused in comparison, rather more relaxed, a little hazy and less detailed (despite its having better signal capacitors and a stepped attenuator).
Side by side, the t-amp sounds superior in many ways but on prolonged listening it is the JLH which makes the better music....somehow!
Tim.
Hi Thijs,
that are only values from a spice simulation, not measurements from a real Circuit !
And it is not yet optimized !
In the single version there are 3 pair MJ11016 output Transistors with 0,9 A bias current per Transistor.
In the brigde version there are 5 pair MJ11016 (per single Amp, complete 10 pair, i know it is overkill ;-) with 0,5 A bias current per Transistor.
And yes, i can post the Schematic. Its attached.
Greetings
Small_boy
that are only values from a spice simulation, not measurements from a real Circuit !
And it is not yet optimized !
In the single version there are 3 pair MJ11016 output Transistors with 0,9 A bias current per Transistor.
In the brigde version there are 5 pair MJ11016 (per single Amp, complete 10 pair, i know it is overkill ;-) with 0,5 A bias current per Transistor.
And yes, i can post the Schematic. Its attached.
Greetings
Small_boy
Attachments
Re: ZVN3310A
ha,thank you very mach,my friend!
but it's too far away from england to china,I don't want to trouble you...thank you again!
last night I have made some similation with ZVN3306A in my MULTISIM2001 software...but it's THD didn't decrease,so I suppose ZVN3306A(3310A) has a high UGS(about 4V),ZVN3306A's UGS add IRF510's UGS is about 8V,let Q1's UCE is reduced,so it's THD increased,that is to say:you can using more higher UCC(more than 36V) to get more lower distortion,But you have to choose ALEX's circuit if you want to increase it's working Voltage...because the SK170 only a small power transistor...
I will make a sample to test can't it works well...just wait me for few days!
best wish!
WINCO
mikelm said:
ha,thank you very mach,my friend!
but it's too far away from england to china,I don't want to trouble you...thank you again!
last night I have made some similation with ZVN3306A in my MULTISIM2001 software...but it's THD didn't decrease,so I suppose ZVN3306A(3310A) has a high UGS(about 4V),ZVN3306A's UGS add IRF510's UGS is about 8V,let Q1's UCE is reduced,so it's THD increased,that is to say:you can using more higher UCC(more than 36V) to get more lower distortion,But you have to choose ALEX's circuit if you want to increase it's working Voltage...because the SK170 only a small power transistor...
I will make a sample to test can't it works well...just wait me for few days!
best wish!
WINCO
