A search with Google is definitely faster.
By the way:
I have thermally coupled Q5,6,7, it is best to plant them in a common aluminum block.
Q10 also needs its own small heat sink, Q3,4 are not only connected in parallel, but also thermally coupled with each other (small common heat sink)!
HBt.
By the way:
I have thermally coupled Q5,6,7, it is best to plant them in a common aluminum block.
Q10 also needs its own small heat sink, Q3,4 are not only connected in parallel, but also thermally coupled with each other (small common heat sink)!
HBt.
The NFB network is drawn on the right with a 9.3 gain.
With a gain that low i would be much concerned by the stability given the rest of the circuitry.
over the thumb
I could not detect any instabilities on the final object.
static:
[(R18+R19) / R19] * [R20/(R20+R21)] = 9.3 * 0.89 = 8.28
18.4dB voltage-gain!
We don't need much more than 2 real watts to enjoy music and our CDP could supply us with 5.66Vpp. I think that should be kept in mind, especially with the loudness -maximized conservation. The input resistance of the 20WPA is frequency -dependent and quite low -resistance!
I could not detect any instabilities on the final object.
static:
[(R18+R19) / R19] * [R20/(R20+R21)] = 9.3 * 0.89 = 8.28
18.4dB voltage-gain!
We don't need much more than 2 real watts to enjoy music and our CDP could supply us with 5.66Vpp. I think that should be kept in mind, especially with the loudness -maximized conservation. The input resistance of the 20WPA is frequency -dependent and quite low -resistance!
Assuming that the parasitic and negative feedback capacitance of the base collector diode path (in reverse bias, of course) for a BD139 is typically 16.22pF (at 10kHz), this results in a 32.44pF parasitic and negative feedback (non-linear, voltage and frequency dependent) capacitance.
Now an ohmic resistor with a value of 13k7 Ohm is connected in parallel and not a 10 times larger ceramic disk capacitor as is normally the case.
From a theoretical and highly simplified point of view, this leads to an fh of this stage, the VAS, of greater than 350kHz.
and so on.
The VAS is a standard, basic emitter circuit with local feedback via R7 (and just a little bit, very tiny, R5,6) - whose input resistance virtually tends to zero.
and so on.
Now an ohmic resistor with a value of 13k7 Ohm is connected in parallel and not a 10 times larger ceramic disk capacitor as is normally the case.
From a theoretical and highly simplified point of view, this leads to an fh of this stage, the VAS, of greater than 350kHz.
and so on.
For me, the circuit topology is an open book and can be read and calculated directly. It's a pity that you obviously want to see it differently - and haven't found the SPICE models yet.
The VAS is a standard, basic emitter circuit with local feedback via R7 (and just a little bit, very tiny, R5,6) - whose input resistance virtually tends to zero.
and so on.
Wow! This is a weird one.
I see three problems:
* The EF1 output stage does not have enough current gain. As a result, the differential and voltage amplifier stages must be run at much higher currents than with an EF2 or EF3.
* The frequency compensation is performed by C2 shunting the input. I suppose the designers can rightfully claim that the amplifier cannot slew because it shorts out any inputs that try to drive it too quickly! I am also not impressed by the use of capacitive voltage dividers C11-C14.
* R7 sets the open-loop gain by coupling ripple from the negative rail. I expect this amplifier to hum.
I believe this is a working amplifier, but it is from a time of not-very-good amplifiers.
Ed
I see three problems:
* The EF1 output stage does not have enough current gain. As a result, the differential and voltage amplifier stages must be run at much higher currents than with an EF2 or EF3.
* The frequency compensation is performed by C2 shunting the input. I suppose the designers can rightfully claim that the amplifier cannot slew because it shorts out any inputs that try to drive it too quickly! I am also not impressed by the use of capacitive voltage dividers C11-C14.
* R7 sets the open-loop gain by coupling ripple from the negative rail. I expect this amplifier to hum.
I believe this is a working amplifier, but it is from a time of not-very-good amplifiers.
Ed
Hi hbt,
Do you have possibility to measure this amp and present results?
Imo that would help discussion, so far seen enough sims that don't represent weird reality, without undermining sims that are mostly good when done right.
Do you have possibility to measure this amp and present results?
Imo that would help discussion, so far seen enough sims that don't represent weird reality, without undermining sims that are mostly good when done right.
Strangely, I don't notice any hum even with my ear directly on the membrane of the loudspeaker.
It is a really weird - And I hesitated for a very long time to introduce this thing here.
HBt.
Subjective , it's just different. And in 1979 .... there were VERY good amplifiers.
79-80 is about when the 25-30mhz Ft Sanken and Toshiba OP's came out. Fast EF2/3's.
Most of the standard forum designs are early 80's with better parts (tweaked with the simulator).
OS
The amplifier's feedback is correcting the ripple-coupling resistor. The hum may be visible on an oscilloscope.
Ed
1979 had both good and bad amplifier designs. Today, one rarely finds bad designs.
Ed
Of course I carried out all the standard measurements back then (for the initial commissioning), the oscilloscope is my favorite eye ..!
Since I built this little amp just for fun and for myself, and never expected to post it anywhere, my documentation was very sloppy.
#
I hope,
the thread is still not boring and a little inspiring. Simulation can go full speed ahead for all I care. I just got a bit tired of the current ABBA thread.
HBt.
.SUBCKT tip142 1 2 3
Q2 1 4 3 q1model 19.6725
D1 3 1 dmodel
R1 2 4 8000
R2 4 3 40
+IS=1e-12 RS=10 N=1 XTI=3
+CJO=0 VJ=0.75 M=0.33 FC=0.5
.MODEL qmodel npn
+IS=2.92836e-16 BF=233.76 NF=1.2 VAF=1000
+IKF=0.222129 ISE=4.61322e-11 NE=2.48158 BR=1.89854
+NR=0.75 VAR=168.458 IKR=0.108495 ISC=9.99999e-14
+NC=2.00001 RB=7.90758 IRB=0.2 RBM=7.90758
+RE=0.275456 RC=1.42279 XTB=0.232595 XTI=3.02293 EG=1.206
+CJE=1e-11 VJE=0.75 MJE=0.33 TF=1e-09
+XTF=1 VTF=10 ITF=0.01 CJC=1.67157e-09
+VJC=0.95 MJC=0.23 XCJC=0.9 FC=0.5
+TR=1e-07 PTF=0 KF=0 AF=1
.MODEL q1model npn
+IS=2.92836e-16 BF=233.76 NF=1.2 VAF=1000
+IKF=0.222129 ISE=4.61322e-11 NE=2.48158 BR=1.89854
+NR=0.75 VAR=168.458 IKR=0.108495 ISC=9.99999e-14
+NC=2.00001 RB=7.90758 IRB=0.2 RBM=7.90758
+RE=0.275456 RC=1.42279 XTB=0.232595 XTI=3.02293 EG=1.206
+CJE=1e-11 VJE=0.75 MJE=0.33 TF=1e-09
+XTF=1 VTF=10 ITF=0.01 CJC=0
+VJC=0.95 MJC=0.23 XCJC=0.9 FC=0.5
+TR=1e-07 PTF=0 KF=0 AF=1
.ENDS
.SUBCKT tip147 1 2 3
Q2 1 4 3 q1model 9.26065
D1 1 3 dmodel
R1 2 4 8000
R2 4 3 40
+IS=1e-12 RS=10 N=1 XTI=3
+CJO=0 VJ=0.75 M=0.33 FC=0.5
.MODEL qmodel pnp
+IS=1.03024e-14 BF=257.093 NF=1.2 VAF=679.126
+IKF=0.190032 ISE=1.23159e-13 NE=1.73997 BR=1.10206
+NR=1.19838 VAR=126.182 IKR=0.103332 ISC=1.23159e-13
+NC=2 RB=5.06065 IRB=0.2 RBM=5.06065
+RE=0.199903 RC=2.01466 XTB=0.354371 XTI=3.01199 EG=1.206
+CJE=1e-11 VJE=0.75 MJE=0.33 TF=1e-09
+XTF=1 VTF=10 ITF=0.01 CJC=1.67157e-09
+VJC=0.95 MJC=0.23 XCJC=0.9 FC=0.5
+TR=1e-07 PTF=0 KF=0 AF=1
.MODEL q1model pnp
+IS=1.03024e-14 BF=257.093 NF=1.2 VAF=679.126
+IKF=0.190032 ISE=1.23159e-13 NE=1.73997 BR=1.10206
+NR=1.19838 VAR=126.182 IKR=0.103332 ISC=1.23159e-13
+NC=2 RB=5.06065 IRB=0.2 RBM=5.06065
+RE=0.199903 RC=2.01466 XTB=0.354371 XTI=3.01199 EG=1.206
+CJE=1e-11 VJE=0.75 MJE=0.33 TF=1e-09
+XTF=1 VTF=10 ITF=0.01 CJC=0
+VJC=0.95 MJC=0.23 XCJC=0.9 FC=0.5
+TR=1e-07 PTF=0 KF=0 AF=1
.ENDS
**************************************
* Model Generated by MODPEX *
Copyright(c) Symmetry Design Systems
- Model generated on Feb 8, 2004
- Model format: PSpice
- Darlington macro model
- External node designations
- Node 1 -> Collect
- Node 2 -> Base
- Node 3 -> Emitter
Q2 1 4 3 q1model 19.6725
D1 3 1 dmodel
R1 2 4 8000
R2 4 3 40
- Default values used in dmodel
- EG=1.11 TT=0 BV=infinite
+IS=1e-12 RS=10 N=1 XTI=3
+CJO=0 VJ=0.75 M=0.33 FC=0.5
.MODEL qmodel npn
+IS=2.92836e-16 BF=233.76 NF=1.2 VAF=1000
+IKF=0.222129 ISE=4.61322e-11 NE=2.48158 BR=1.89854
+NR=0.75 VAR=168.458 IKR=0.108495 ISC=9.99999e-14
+NC=2.00001 RB=7.90758 IRB=0.2 RBM=7.90758
+RE=0.275456 RC=1.42279 XTB=0.232595 XTI=3.02293 EG=1.206
+CJE=1e-11 VJE=0.75 MJE=0.33 TF=1e-09
+XTF=1 VTF=10 ITF=0.01 CJC=1.67157e-09
+VJC=0.95 MJC=0.23 XCJC=0.9 FC=0.5
+TR=1e-07 PTF=0 KF=0 AF=1
.MODEL q1model npn
+IS=2.92836e-16 BF=233.76 NF=1.2 VAF=1000
+IKF=0.222129 ISE=4.61322e-11 NE=2.48158 BR=1.89854
+NR=0.75 VAR=168.458 IKR=0.108495 ISC=9.99999e-14
+NC=2.00001 RB=7.90758 IRB=0.2 RBM=7.90758
+RE=0.275456 RC=1.42279 XTB=0.232595 XTI=3.02293 EG=1.206
+CJE=1e-11 VJE=0.75 MJE=0.33 TF=1e-09
+XTF=1 VTF=10 ITF=0.01 CJC=0
+VJC=0.95 MJC=0.23 XCJC=0.9 FC=0.5
+TR=1e-07 PTF=0 KF=0 AF=1
.ENDS
.SUBCKT tip147 1 2 3
- Model generated on Feb 8, 2004
- Model format: PSpice
- Darlington macro model
- External node designations
- Node 1 -> Collect
- Node 2 -> Base
- Node 3 -> Emitter
Q2 1 4 3 q1model 9.26065
D1 1 3 dmodel
R1 2 4 8000
R2 4 3 40
- Default values used in dmodel
- EG=1.11 TT=0 BV=infinite
+IS=1e-12 RS=10 N=1 XTI=3
+CJO=0 VJ=0.75 M=0.33 FC=0.5
.MODEL qmodel pnp
+IS=1.03024e-14 BF=257.093 NF=1.2 VAF=679.126
+IKF=0.190032 ISE=1.23159e-13 NE=1.73997 BR=1.10206
+NR=1.19838 VAR=126.182 IKR=0.103332 ISC=1.23159e-13
+NC=2 RB=5.06065 IRB=0.2 RBM=5.06065
+RE=0.199903 RC=2.01466 XTB=0.354371 XTI=3.01199 EG=1.206
+CJE=1e-11 VJE=0.75 MJE=0.33 TF=1e-09
+XTF=1 VTF=10 ITF=0.01 CJC=1.67157e-09
+VJC=0.95 MJC=0.23 XCJC=0.9 FC=0.5
+TR=1e-07 PTF=0 KF=0 AF=1
.MODEL q1model pnp
+IS=1.03024e-14 BF=257.093 NF=1.2 VAF=679.126
+IKF=0.190032 ISE=1.23159e-13 NE=1.73997 BR=1.10206
+NR=1.19838 VAR=126.182 IKR=0.103332 ISC=1.23159e-13
+NC=2 RB=5.06065 IRB=0.2 RBM=5.06065
+RE=0.199903 RC=2.01466 XTB=0.354371 XTI=3.01199 EG=1.206
+CJE=1e-11 VJE=0.75 MJE=0.33 TF=1e-09
+XTF=1 VTF=10 ITF=0.01 CJC=0
+VJC=0.95 MJC=0.23 XCJC=0.9 FC=0.5
+TR=1e-07 PTF=0 KF=0 AF=1
.ENDS
**************************************
* Model Generated by MODPEX *
Copyright(c) Symmetry Design Systems
- All Rights Reserved *
- UNPUBLISHED LICENSED SOFTWARE *
- Contains Proprietary Information *
- Which is The Property of *
- SYMMETRY OR ITS LICENSORS *
- Modeling services provided by *
- Interface Technologies www.i-t.com *
Output devices are darlingtons, but this change nothing to the weirdness of this amp.
I'm slowly losing the confidence to report on this really strange design, because I have to post from memory.
But:
I have (had) compared the distortion spectrum with that of the classic SYMM (both subjects on the table) - but the FFT did not show any higher (order) or dissonant components. As I said, I don't dare to say it.
And yes, 15mA | 32mA | 2 times >30mA is a really strangely high number to begin with. I admit that. All this was the reason why I built this circuit without further hesitation and, above all, thought.
I still have a small photo gallery on some USB stick, which I will be happy to look for if that should help.
Good Night,
HBt.
But:
I have (had) compared the distortion spectrum with that of the classic SYMM (both subjects on the table) - but the FFT did not show any higher (order) or dissonant components. As I said, I don't dare to say it.
And yes, 15mA | 32mA | 2 times >30mA is a really strangely high number to begin with. I admit that. All this was the reason why I built this circuit without further hesitation and, above all, thought.
I still have a small photo gallery on some USB stick, which I will be happy to look for if that should help.
Good Night,
HBt.
Those old darlingtons can have quite low gain at high current but that doesnt justify a VAS current that huge,
that s just designs based on electronics dedicated urban legends.
There was a belief at the time that previous stages being heavily class A would imprvove the sound quality,
wich we know is wrong, as another exemple Bob Cordell s BC1 has a VAS that drain 10mA and yet it is loaded
by an EF3 that has at least a 500 000 gain, even its EF3 is weird since the pre drivers are at 10mA and the drivers at 27mA,
and since those two stages are not referenced to the output those high currents dont even help to reduce the crossover THD.
It’s only a +/-25V supply, meaning 3 or 4 amps peak for each 8 ohm load attached. The gain is high enough on these at 4A.
Biggest problem with these darlingtons is the fact that you’re stuck with a 3 MHz driver transistor. You really want a high fT, sustained beta device in that position, even if you are using old-and-slow outputs. Even the old classic TO-5 metal can parts were pretty nice in that regard, and the Japanese seem to have perfected them in the 70’s and 80’s - then stuck them in plastic packages and sold them for under a buck. It’s way better than no driver, as @saabracer23 found out recently, but with proper drivers the same architecture is better. It’s too bad the old console he was restoring didn’t have a place to put them.
Biggest problem with these darlingtons is the fact that you’re stuck with a 3 MHz driver transistor. You really want a high fT, sustained beta device in that position, even if you are using old-and-slow outputs. Even the old classic TO-5 metal can parts were pretty nice in that regard, and the Japanese seem to have perfected them in the 70’s and 80’s - then stuck them in plastic packages and sold them for under a buck. It’s way better than no driver, as @saabracer23 found out recently, but with proper drivers the same architecture is better. It’s too bad the old console he was restoring didn’t have a place to put them.