You will always put more load on the VAS doing this, even using paralleled drivers. The gain is the same. You just don’t want to run the drivers too high in current, because their gain will drop even further. The solution is a driver designed for the larger current (and dissipation) so the beta stays high. You don’t want the added stability risk (Including thermal stability issues) of using individual drivers. If they don’t run at the same temperature you will have trouble with it all. The “additional load” on the VAS will result in lower open loop gain when loaded at 4 ohms. I suppose you could turn it into an EF3, but that would require a lot more of a kluge on the board and might give unanticipated problems. I would increase the VAS current first before resorting to that, if it indeed does run out of current. Bigger heat sink may be required, perhaps upgraded transistor.
Individual emitter resistors goes without saying…..
I have a bunch of nwj0302's, thinking of using these are drivers, gain is quite stable even between 1-3amps
1) Remove R36. Connect the base Q18 with a 10 ohm resistor and a wire to the left terminal R35.
2) Remove R38. Connect base Q16 with a 10 ohm resistor and a wire to the left terminal R37.
3) Connect the outputs (after the contacts of the protection relay?). On external output terminals.?
Set the quiescent current (bias) to minimum.
2) Remove R38. Connect base Q16 with a 10 ohm resistor and a wire to the left terminal R37.
3) Connect the outputs (after the contacts of the protection relay?). On external output terminals.?
Set the quiescent current (bias) to minimum.
Last edited:
Much to the delight of the naysayers, I have run into my first road-block. Running a few tests to see if I can get the existing set up to give me 40v or so with high impedances as a proof of concept.
Seems the protection circuit is overly eager. It clips the signal even when the current limit on the outputs is not reached. eg with 8 ohms it clips at about 34.5v, I then tested with a 30 ohm load, it does the same, clips at 34.5v also, apparently its not only current limiting, it is voltage clamping as well.. With the rails staying over 75vdc, I would have expected clipping at 40vac or over. I am now considering removing all the protection transistors (Q23 to Q30) otherwise this project will not add much value.
Will keep you guys updated.
Seems the protection circuit is overly eager. It clips the signal even when the current limit on the outputs is not reached. eg with 8 ohms it clips at about 34.5v, I then tested with a 30 ohm load, it does the same, clips at 34.5v also, apparently its not only current limiting, it is voltage clamping as well.. With the rails staying over 75vdc, I would have expected clipping at 40vac or over. I am now considering removing all the protection transistors (Q23 to Q30) otherwise this project will not add much value.
Will keep you guys updated.
Add to that doubling Vas idle current.
Input differential pair will probably not need any modding.
Please don´t be mistaken, no idle "naysayers" here, rather "been there done that before" types. 😊
Yes, that might be an issue, so I will just take the output from one channel VAS.
To me, the difference between been there done that and naysayers is, the Naysayers gives reasons not to do something, the other guys tell you what you can do. I have seen responses of some people on the forum, I never doubt patterns of responses even if one gives some benefit of doubt for one response.... ofcourse human behavior is quite complex...
The protection circuit needs to be appropriately “scaled” to limit at 2x the current. Sometimes easy, sometimes hard. In the case where it senses the voltage across the emitter resistor, it sort of does that automatically if you put an extra transistor in parallel with another of the same emitter resistor.
Naysayers? No, I’ve added extra output transistors to amplifiers which don’t like 4 ohm loads before. I think of it is “it should have been that way in the first place but they we’re too cheap to do it”.
Naysayers? No, I’ve added extra output transistors to amplifiers which don’t like 4 ohm loads before. I think of it is “it should have been that way in the first place but they we’re too cheap to do it”.
Exactly.
for the current limiting, for some reason it was voltage limiting too. ie with doubled outputs, while I was expecting 300/4e, in 8 ohms, I was hoping for 200w/8e, the protect circuit was limiting it. now it swings all the way to 46vrms, up from 34.5vrms.
Just finished one of the 2 amps.
- I removed the outputs and stuck in 4 pairs of 2sa1943/2sc5200's. I gain and vbe matched them to less than 4% variation
- All PNP's are on one heatsink, and NPN's on the other. I did this for closer coupling and better thermal tracking.
- Outputs run via 0.25Ohm ER's
- Took the input for the outputs via 27 ohm base resistors tied to the drivers. Used drivers from left channel to drive all 8 output devices.
- At 300Hz, the waveform clipped at an unexpected 42.3vrms into my 4 ohm woofer. The impedance might be higher, will use fixed resistors tomorrow. I did not expect this much power from that smallish transformer. The sound does feel slightly strained at the highest volumes, but thats to be expected in an "elastic" PSU.
- Had to jack up the bias from stock, currently keeping it at 36ma due to smallish heatsinks. Might pull it back a bit till I start to see crossover distortion on the scope. But this is very close to what the Service manual suggests.
- There is no DC offset pot, so there is about 26mv of DC on the outputs, but the protection does not complain. Might be ok.
- Drivers getting warm to touch (40c?) these are stock.... wondering if I should put larger ones in the or mount these to the heatsink.
- There is no thermal cut out on this amp, so apart from the vbias mounted on the heatsink, nothing tracking overheating. I might install thermal breakers later on.
- Considering changing the input coupling caps to film types, it currently has 40 year old electrolytics. That could be a cause of the 2kHz - 3khz brightness I was hearing at high volumes.
- The drivers of the "unused" channel got very very hot, so I removed them from the board. Hope nothing else is cooking...
- The amplifier has no DC rail fuses, this is something I need to add later on.
Last edited:
Congratulations on your partial success ! Schematic of what you did when you have time please!
I am curios 81V drops to ? at maximum power (before clipping) ?
Now I have noticed the 2 channels are horizontal mirrored in the schematic - I am sorry yesterday it was late in the night and the resolution was low I didn't notice the sense of arrows on the emitters, I taught they are identical one on top of each other (not mirrored); the green drawing is wrong, the correct way is base of Q22 with base of Q21 (both PNP) and base of Q20 with base of Q19 (both NPN) I hope you got that right...
We kind of expected that, remember you said about them "if they survive"; more heatsink is needed for them
It really depends on the amplification factor of the transistors you have; some calculations:
The Q8 Q11 Ivas current is given by (Supply Voltage - Vbe17 - Vbe21 - Vcp1) / (R33 + R31)
Because of the bootstrap connection C15 voltage on R31 and the current going through are (almost) constant
Considering the worst case scenario, at maximum power on the speaker (maybe at 10A ? current if you say it clips around 42.3 V on 4 ohm) the supply voltage drops from 81V to (maybe ?) 68V
Ivas = (68V - 0.7V - 0.7V - 0.22ohm x (10A/number of parallel transistors =2)) / (9.1K + 8.2K) = 3.79 mA
Compare this to the max curent required in base of Q17 simplified formula ib17 = ie17 / h17 =~ ib21 / h17 = (ie21/h21) / h17
for example if your Q17 2SA1535A has h17 = 100 and your Q21 2SA1216 has h21=60 you will get
Ib17max = (10A/60)/100 = 1.67 mA Notice that if this value is lower than Ivas 3.79 mA (better with some reserve) , then you are all set, however for lower h17 or/and h21 you might be not (Ib17max > Ivas). If that is the case you need to increase Ivas by reducing R33 and R31 or you need transistors with higher h.
Last edited:
Floriant, Thank you SO much for your going over the circuit and explaining the gain in this scenario. I will reply to each of your questions later in the day once I go down to my work area.
For now, Here is what I kludged in photoshop to show you what I did.
I did remove
Installed the following:
For now, Here is what I kludged in photoshop to show you what I did.
I did remove
- All protection transistors from both channels (Q23-30?), and
- both drivers from the non working channel.
- Also removed original outputs from both sides, replaced with Toshibas.
Installed the following:
- 2sc5200 x4 (hfe between 96-100)
- 2sa1943 x4 (hfe between 88-96)
- 0.25R 5w x8 (This is up from 0.22 in stock)
- 27R 1/4w x8 (What I had in hand in quantity)
Are Q15 and 17 TO-220 or TO-126 types? If they are the latter, they will be at most risk of becoming the next victims. They need to be heat sinked for at least 3.2 watts continuous dissipation each, and good for 16 watts @ 40V or more. Most of the TO-220 Jap drivers and the MJE1503x are good for this, but not all the smaller ones are. Heat sinking for 3+ watts is easier with metal tab (not plastic over molded) TO-220’s.
I had an EF3 amp running on +/-75, with 5 pairs of TO-3 outputs, TO-3 drivers, and Sanyo TO-220 predrivers. One channel had a mistake on the PCB, shorting base-emitter on the driver. That channel “biased funny” - as in a different setting on the Vbe pot to get the current right. But it ran fine in PA duty for two years, until the “driver“ transistor (which was supposed to be a predriver) blew. It didnt have a heat sink because it wasn’t supposed to need one. I know your outputs have a higher beta the the MJ15022, but dont expect miracles.
I had an EF3 amp running on +/-75, with 5 pairs of TO-3 outputs, TO-3 drivers, and Sanyo TO-220 predrivers. One channel had a mistake on the PCB, shorting base-emitter on the driver. That channel “biased funny” - as in a different setting on the Vbe pot to get the current right. But it ran fine in PA duty for two years, until the “driver“ transistor (which was supposed to be a predriver) blew. It didnt have a heat sink because it wasn’t supposed to need one. I know your outputs have a higher beta the the MJ15022, but dont expect miracles.
Ran some more reliable tests today:
The US version apparently has lower supplies ie 54-0-54vac Transformers. After rectification, the dc rails are about +/- 72.1vdc
@ 4 ohms with a 200Hz sinewave: Output clipped at 39.6vrms. The rails sagged from 72 to 59.3vdc
@ 8 ohms with a 200Hz sinewave: Output clipped at 41.2 vrms. The DC rails sagged to about 64.3vdc.
So this transformer might be slightly better regulated than the international version which has dual supplies of 58-0-58 and 34-0-34 for 8 and 4 ohm operation. This one has a single 54-0-54 supply. I tried to size up the transformer, it has the exact dimensions as the Kenwood Basic M2a, which uses 2 of these transformers and is rated for 300wpc+ into 4 ohms. The laminate core is about 2" x 4" x 4.2" as is this guy.
What do you think the loading on the VAS will be for this scenario? IS my math correct?
Ivas = ((59.3 - 1.45v x (10A/4 =2.5)) / 17.3k = 8.35mA ??? something is off.
You are welcome ! Thank you for sharing results and schematic !
You know how people have different ideas at different moments in life ? One of mine was related to all the receivers that get thrown away in the landfill (defective or not) because of the obsolete DSP or digital features while the audio power amplifier part in receivers let say sold in the last 10 or maybe more years are as good (except maybe some electrolitic capacitors) in terms of performance and schematic as a new one. So my ideea was to transform the 5.1 or 7.1 (digital obsolete) receivers in mono 300 - 700 W amplifiers with just input and output and give them a new life. Even if I never really pursued the idea (I was not sure if anyone will be interested in buying those) I have thought about this process before.
Now that you have provided the values we can calculate; I wrote the formula as
Ivas = (Supply Voltage - Vbe17 - Vbe21 - Vcp1) / (R33 + R31) so
a) with no signal (_)
Vcp1 = 40mA x 0.25 ohm = 10mV
Ivas_ = (72V - 0.7V - 0.7V - 0.01) / 17.3 kohm = 4.196mA
b) at max power on 4 ohm (M4) you say V clip 39.6vrms that's 55.8 Vpeak
peak current trough speaker I= U/R = 55.8 / 4 = 13.95 A
Divided trough the 4 Q21 transistors, each get 3.49A
Vcp1 = 3.49A x 0.25 ohm = 0.87V
Considering a hq21=88, Ib21 = 3.49A / 88 = 39 mA
Voltage on 27 ohm resistor = 27 ohm x 39mA = 1V (this is resistor voltage drop as CP1 ) so
IvasM4 = (59.3V - 0.7V - 0.7V - 0.87V - 1V) / 17.3 kohm = 3.2mA
b) is an approximation of worst case scenario (because h could be different and changes with the current, at 3.49A Vbe is probably something like 0.8V instead of 0.7V used in formulas, there is one more 10 ohm resistor in Q17 base, some millivolts on that, and so on).
I will attach an excel file with calculations; fields with border around are user inputs.
Optional, with low importance: You can remove all transistors from unused channel ("nothing else will be cooking") and you can increase bootstrap capacitor C15; if you don't have any, remove and put C16 (from the unused channel) in paralel on C15; attention to polarity!