I recently started experimenting with a 2N2222 transistor to amplify the output of a MK-484 kit in hopes of driving an Edcor WSM15K/15K transformer to drive a push pull 60FX5 amp.
That didn't work so well so I looked for other transistors I could use and experimented with the 2N6301 which is a darlington power transistor.
Got a basic circuit worked up with one transistor then thought why not try push pull.
Through experimentation I came up with this circuit at the bottom of the post.
This is to drive a high impedance 1920's speaker.
I adjusted bias until I got a good clean sinewave then as the transistors warmed up I kept adjusting bias until I got the maximum possible voltage with a clean sinewave which is 60Vrms (169.68Vpp) on the output transformer secondary using a 5K load, but because the transformer is only rated to 50Vrms (141.4Vpp) I did all my tests at that output voltage.
I tested the amp using a 5K to 8 ohm transformer and an Automatic Radio speaker I modded to be three way using a dome tweeter and the amp itself does sound quite good.
Driving the antique speaker the amp sounds quite decent as well and drives the speaker quite nicely.
I used a 15 volt zener pressed against the small chassis between where the two transistors are mounted to serve as temperature compensation, but even though it works some it doesn't work well enough.
When I have everything set up so that I get a good clean sinewave of 50Vrms out of the amp I can remove B+ then let the amp cool and when I power it up again the output voltage starts at 35Vrms and as the amp warms up goes to 50Vrms.
The capacitor is only there to eliminate a bit of instability when bias was set to a certain point.
Any idea what I can do to have the bias properly compensate for temperature so that the amp will always have maximum gain regardless of the temperature of the transistors?
The transistors themselves don't get much above warm to the touch.
I find it amazing that with only two transistors and two transformers I can get a 50Vrms output for an input of only 40.55mVrms.
That didn't work so well so I looked for other transistors I could use and experimented with the 2N6301 which is a darlington power transistor.
Got a basic circuit worked up with one transistor then thought why not try push pull.
Through experimentation I came up with this circuit at the bottom of the post.
This is to drive a high impedance 1920's speaker.
I adjusted bias until I got a good clean sinewave then as the transistors warmed up I kept adjusting bias until I got the maximum possible voltage with a clean sinewave which is 60Vrms (169.68Vpp) on the output transformer secondary using a 5K load, but because the transformer is only rated to 50Vrms (141.4Vpp) I did all my tests at that output voltage.
I tested the amp using a 5K to 8 ohm transformer and an Automatic Radio speaker I modded to be three way using a dome tweeter and the amp itself does sound quite good.
Driving the antique speaker the amp sounds quite decent as well and drives the speaker quite nicely.
I used a 15 volt zener pressed against the small chassis between where the two transistors are mounted to serve as temperature compensation, but even though it works some it doesn't work well enough.
When I have everything set up so that I get a good clean sinewave of 50Vrms out of the amp I can remove B+ then let the amp cool and when I power it up again the output voltage starts at 35Vrms and as the amp warms up goes to 50Vrms.
The capacitor is only there to eliminate a bit of instability when bias was set to a certain point.
Any idea what I can do to have the bias properly compensate for temperature so that the amp will always have maximum gain regardless of the temperature of the transistors?
The transistors themselves don't get much above warm to the touch.
I find it amazing that with only two transistors and two transformers I can get a 50Vrms output for an input of only 40.55mVrms.
Attachments
Question asked:
> have the bias properly compensate for temperature
The Zener is NOT temperature correcting, it has very low temperature coefficient.
It also is not setting the Base ("grid") voltage relative to Emitter (cathode), it is feeding a fraction of B+ through a high value pot.
Since the transistors are not simple (Darlingtons, and also have BE resistors of unknown tempco), analysis is screwy. It resembles the old-old-old scheme of leaking a constant current into Base, and hoping that hFE doesn't change (it does). That is so 1953.
One transistor's Vbe varies 0.5V-0.8V, use 0.6V for rough guesses. It varies with current and with temperature, so you can't give (or work to) an exact figure. What to do? Add an emitter resistor (you got that) which will drop a voltage "large" in comparison to the uncertainty of Vbe. A 0.6V drop in emitter resistor is usually safe, since the actual variation of Vbe in a specific device is rarely over 0.1V and we rarely need better than 20% accuracy of Ie. (Your no-feedback circuit's gain will vary about as Ie, and while 20% is not good enough for recording consoles it is fine for a radio.)
But what current should we want? Look at the supply and load. 24V supply, 600r CT load. Assume class A. Each side reflects 300 Ohms. Assume much of the 24V is to be applied to the winding as peak audio. 21V peak seems a nice number. Especially since the RMS of 21V sine is 15V. Or 30Vrms across the whole 600 Ohm winding. 30V^2/600 is 1.5 Watts output. This seems a lot, but would make sense for a line driver.
The 21V peak in 300r is 70mA peak. In class A we would idle at 35mA, each side. 70mA idle for both sides.
We want a emitter resistor to drop ~~0.6V at 70mA. Comes out "10 Ohms" on my thumbs. Interesting that this is the value you picked.
We want the Bases to sit 0.6V+0.6V= 1.2V higher than the ~~0.7V at emitter. We want this ~~1.8V to be steady with small supply changes (no resistor divider) and with base current change (no large resistance), and preferably drop with temperature. Well, Vbe does that naturally. If we only want a Vbe, we don't need the whole transistor, a diode is a transistor without the third leg. To get ~~1.8V we need three diodes.
The diode current must be enough to get the diode voltage up, plus enough to feed the base currents, plus any other stuff we may steal.
The base current of a 2N6301 at 35mA or 0.035A..... well, it is a 8 AMP part with no low-current specs. We might wonder if it is even "alive" at 1% of its marketed current rating. (Old Ge might not. Modern Si usually will. Your tests show it ain't dead at lower current.)
We don't know the hFE of a '6301 at 35mA. However some poking shows that the 8K resistor across the first Vbe will suck at least 0.070mA. And if the hFE of each transistor is as low as 20(!), the base current will be no large. Pencil "0.1mA?" as base current. Times two is 0.2mA.
Take 1mA as a minimum current to bring super-cheap reliable 1N400x diodes up to the 0.6V zone. (Note that this is only 0.1% of '400x rating! I know from other work that it will be a decent bias source.)
And since the expected 1.8V is remarkably close to that that MK484 eats, and it only sips 0.3mA, figure on that.
0.2mA+1mA+0.3mA is 1.5mA. We need a resistor for (24V-1.8V)/1.5mA, which is like 15K. No harm in using 10K.
The internal dynamic resistance of a Si diode at 1mA is 26 Ohms (Shockley's Law). Three stacked is 78 Ohms. This may be a low-low impedance for the bases. In fact it is likely far lower than transformer winding resistance.
DC design is done.
Computed power output seems absurd to drive a couple of grids?
Gain can be computed several ways. Simplest is to know that 20mV at the B-E of a transistor will cause a very large change of current and perhaps near-maximum output. This reflects to 40mV peak at the input transformer. 40mV is round-about the output of the MK484.
Alternatively Shockley's Law teaches that at 35mA the transistor emitter dynamic impedance is 0.74 Ohms, from there and the load work out the voltage gain.
> 50Vrms output for an input of only 40.55mVrms.
See? The flip side is low input Z and high distortion. Transistor amps "always" need significant NFB. This may be the marginal case.
So we can just about drive to clipping. Clipping output is 30Vrms in 600r or 60Vrms at the 2,400r winding. This seems somewhat higher than the 60FX5's need for 2Vrms drive (and 500K input loading).
In fact it seems over-kill for cost, gain, power in, power out. If the OT were 600CT:8 and any decent common speaker attached, it would alone wake the whole house and compete with any two-60FX5 amp.
> have the bias properly compensate for temperature
The Zener is NOT temperature correcting, it has very low temperature coefficient.
It also is not setting the Base ("grid") voltage relative to Emitter (cathode), it is feeding a fraction of B+ through a high value pot.
Since the transistors are not simple (Darlingtons, and also have BE resistors of unknown tempco), analysis is screwy. It resembles the old-old-old scheme of leaking a constant current into Base, and hoping that hFE doesn't change (it does). That is so 1953.
One transistor's Vbe varies 0.5V-0.8V, use 0.6V for rough guesses. It varies with current and with temperature, so you can't give (or work to) an exact figure. What to do? Add an emitter resistor (you got that) which will drop a voltage "large" in comparison to the uncertainty of Vbe. A 0.6V drop in emitter resistor is usually safe, since the actual variation of Vbe in a specific device is rarely over 0.1V and we rarely need better than 20% accuracy of Ie. (Your no-feedback circuit's gain will vary about as Ie, and while 20% is not good enough for recording consoles it is fine for a radio.)
But what current should we want? Look at the supply and load. 24V supply, 600r CT load. Assume class A. Each side reflects 300 Ohms. Assume much of the 24V is to be applied to the winding as peak audio. 21V peak seems a nice number. Especially since the RMS of 21V sine is 15V. Or 30Vrms across the whole 600 Ohm winding. 30V^2/600 is 1.5 Watts output. This seems a lot, but would make sense for a line driver.
The 21V peak in 300r is 70mA peak. In class A we would idle at 35mA, each side. 70mA idle for both sides.
We want a emitter resistor to drop ~~0.6V at 70mA. Comes out "10 Ohms" on my thumbs. Interesting that this is the value you picked.
We want the Bases to sit 0.6V+0.6V= 1.2V higher than the ~~0.7V at emitter. We want this ~~1.8V to be steady with small supply changes (no resistor divider) and with base current change (no large resistance), and preferably drop with temperature. Well, Vbe does that naturally. If we only want a Vbe, we don't need the whole transistor, a diode is a transistor without the third leg. To get ~~1.8V we need three diodes.
The diode current must be enough to get the diode voltage up, plus enough to feed the base currents, plus any other stuff we may steal.
The base current of a 2N6301 at 35mA or 0.035A..... well, it is a 8 AMP part with no low-current specs. We might wonder if it is even "alive" at 1% of its marketed current rating. (Old Ge might not. Modern Si usually will. Your tests show it ain't dead at lower current.)
We don't know the hFE of a '6301 at 35mA. However some poking shows that the 8K resistor across the first Vbe will suck at least 0.070mA. And if the hFE of each transistor is as low as 20(!), the base current will be no large. Pencil "0.1mA?" as base current. Times two is 0.2mA.
Take 1mA as a minimum current to bring super-cheap reliable 1N400x diodes up to the 0.6V zone. (Note that this is only 0.1% of '400x rating! I know from other work that it will be a decent bias source.)
And since the expected 1.8V is remarkably close to that that MK484 eats, and it only sips 0.3mA, figure on that.
0.2mA+1mA+0.3mA is 1.5mA. We need a resistor for (24V-1.8V)/1.5mA, which is like 15K. No harm in using 10K.
The internal dynamic resistance of a Si diode at 1mA is 26 Ohms (Shockley's Law). Three stacked is 78 Ohms. This may be a low-low impedance for the bases. In fact it is likely far lower than transformer winding resistance.
DC design is done.
Computed power output seems absurd to drive a couple of grids?
Gain can be computed several ways. Simplest is to know that 20mV at the B-E of a transistor will cause a very large change of current and perhaps near-maximum output. This reflects to 40mV peak at the input transformer. 40mV is round-about the output of the MK484.
Alternatively Shockley's Law teaches that at 35mA the transistor emitter dynamic impedance is 0.74 Ohms, from there and the load work out the voltage gain.
> 50Vrms output for an input of only 40.55mVrms.
See? The flip side is low input Z and high distortion. Transistor amps "always" need significant NFB. This may be the marginal case.
So we can just about drive to clipping. Clipping output is 30Vrms in 600r or 60Vrms at the 2,400r winding. This seems somewhat higher than the 60FX5's need for 2Vrms drive (and 500K input loading).
In fact it seems over-kill for cost, gain, power in, power out. If the OT were 600CT:8 and any decent common speaker attached, it would alone wake the whole house and compete with any two-60FX5 amp.
Attachments
me> it seems over-kill
I've been at this too long. I hate to see so many parts and so much power when the job can be done so much cheaper.
60FX5 needs 3V peak or 2Vrms of G1 drive to go wild. Super-small for a power tube. 30V 1.2K drive seems absurd.
Push-pull output does not require push-pull input. A long-tail can take one in and give two outs. A disadvantage is you need twice the drive signal. But at only 2Vrms drive, that seems do-able.
This is also called "self-split push-pull". Purists will argue that it tends to PUSH-pull, unbalanced. Large tail resistor improves balance but reduces voltage available to make power output. There is an analysis around. The proposed 100-200r will be good enough. The "optimum" value is whatever is handy in that range.
The input impedance is over 470K due to bootstrapping. We want the previous stage plate (I mean Collector) resistor to be 1/2 to 1/10th that value for good gain and output. I slapped 100K in there.
The bias system looks like the "old-old-old" scheme but isn't. Taking base resistor from collector DC resistor is a form of auto-bias. If the collector sucks down, the base resistor passes less current, so collector doesn't suck so much. A fairly workable Rb for "any" modern small transistor may be 10Meg.
Gain required is 4Vrms/50mVrms or 80. The 1K under the emitter has no DC function, but sets the AC gain near the required value. This can be reduced to zero for more gain, but at 4Vrms out the 2nd harmonic approaches 13%, which isn't why you went push-pull in the final. 1K makes it under 5% 2nd which is probably fine. Trim to taste.
I have not addressed how you are going to get the several supply voltages. Actually you could change the 100K at Tr1 to near 30K, change 1K to 3.3K, feed this with +120V, and still have a safe voltage on Tr1. Now Tr1 output and THD will be even better.
I've been at this too long. I hate to see so many parts and so much power when the job can be done so much cheaper.
60FX5 needs 3V peak or 2Vrms of G1 drive to go wild. Super-small for a power tube. 30V 1.2K drive seems absurd.
Push-pull output does not require push-pull input. A long-tail can take one in and give two outs. A disadvantage is you need twice the drive signal. But at only 2Vrms drive, that seems do-able.
This is also called "self-split push-pull". Purists will argue that it tends to PUSH-pull, unbalanced. Large tail resistor improves balance but reduces voltage available to make power output. There is an analysis around. The proposed 100-200r will be good enough. The "optimum" value is whatever is handy in that range.
The input impedance is over 470K due to bootstrapping. We want the previous stage plate (I mean Collector) resistor to be 1/2 to 1/10th that value for good gain and output. I slapped 100K in there.
The bias system looks like the "old-old-old" scheme but isn't. Taking base resistor from collector DC resistor is a form of auto-bias. If the collector sucks down, the base resistor passes less current, so collector doesn't suck so much. A fairly workable Rb for "any" modern small transistor may be 10Meg.
Gain required is 4Vrms/50mVrms or 80. The 1K under the emitter has no DC function, but sets the AC gain near the required value. This can be reduced to zero for more gain, but at 4Vrms out the 2nd harmonic approaches 13%, which isn't why you went push-pull in the final. 1K makes it under 5% 2nd which is probably fine. Trim to taste.
I have not addressed how you are going to get the several supply voltages. Actually you could change the 100K at Tr1 to near 30K, change 1K to 3.3K, feed this with +120V, and still have a safe voltage on Tr1. Now Tr1 output and THD will be even better.
Attachments
Last edited:
I originally wanted to drive the 60FX5 tubes but then I went through a phase of not knowing what I wanted to do and at that time I remembered I have the Peerless reproducer which is a relatively high impedance driver.
I then designed the circuit based on that driver*and chose the 5K load for testing as that is the driver's impedance at 100Hz which is*the lowest frequency it can reasonably reproduce.
At first I was concerned that the output voltage was a bit too high, but after trying it on the speaker*the output voltage is high enough to give room filling volume.
I can take some measurements when I get home of the current bias voltage..
Will look at those schematics when I get home.
I then designed the circuit based on that driver*and chose the 5K load for testing as that is the driver's impedance at 100Hz which is*the lowest frequency it can reasonably reproduce.
At first I was concerned that the output voltage was a bit too high, but after trying it on the speaker*the output voltage is high enough to give room filling volume.
I can take some measurements when I get home of the current bias voltage..
Will look at those schematics when I get home.
I have had the amp on and sitting idle for at least 20 minutes.
Bias at the center tap of the secondary of the input transformer is 2.282Vdc.
I cannot get the B+ stable enough to measure what the bias voltage is on either side of the secondary given I am not using a regulated supply at the moment, but the bias voltage drops by a few milivolts and one side is slightly lower than the other which I suppose may be due to the winding resistance of the transformer.
B+ current at idle is 100mA.
Also one transistor if I touch it will cause an oscillation in the speaker.
I seem to remember trying the diode idea in some form early on and I could not get gain to remain constant.
That's the goal. To have the bias to where the maximum AC voltage is output and the gain stays the same as the temperature changes.
I can try the diode mod tomorrow at work.
Will three 1N4002 diodes work just as good?
Also would using a voltage regulator to set bias be of any benefit?
The self bias idea seems interesting to me, but I don't quite know how to implement it unless I capacitor couple the input transformer secondary to the transistor bases.
That would add two caps and two resistors and reduce gain somewhat, but it would also be a bit of feedback which might make the amp sound better.
Oh I see so that's 42Vrms maximum across the full 600 ohm primary, right? If so that would be 84Vrms out of the secondary given the transformer has a 1:2 voltage ratio. Does having the bias set where it will be reduce gain? If so I will see if I can get to 50Vrms out of the transformer for the same input voltage. If I cannot then would a 2N2222A work to provide enough gain?
I will need to ask Edcor just what the 50Vrms voltage limit is for the 600 ohm to 2.4K transformer as I don't know if that is just the limitation of one winding or the maximum voltage that can be applied to or output of either winding.
If the 50Vrms limit is just for the 600 ohm winding then I can indeed have much more output voltage using the diodes and a different bias point. That would also reduce the chance of instability and I may then not need the .0033uF cap.
When I get the amp finished I will be using a LM-317 set for 24Vdc to provide the B+ voltage. The MK-484 kit I have ordered has its own voltage divider as it also has a chip amp on it that does at least a watt of power and the circuit can run at up to 9Vdc I think so I could use a 7809 after the 24 volt regulator to provide its voltage. I could have just used that and an 8 ohm to 5K transformer to properly drive the speaker, but I don't think that's necessarily the best way to go.
Bias at the center tap of the secondary of the input transformer is 2.282Vdc.
I cannot get the B+ stable enough to measure what the bias voltage is on either side of the secondary given I am not using a regulated supply at the moment, but the bias voltage drops by a few milivolts and one side is slightly lower than the other which I suppose may be due to the winding resistance of the transformer.
B+ current at idle is 100mA.
Also one transistor if I touch it will cause an oscillation in the speaker.
I seem to remember trying the diode idea in some form early on and I could not get gain to remain constant.
That's the goal. To have the bias to where the maximum AC voltage is output and the gain stays the same as the temperature changes.
I can try the diode mod tomorrow at work.
Will three 1N4002 diodes work just as good?
Also would using a voltage regulator to set bias be of any benefit?
The self bias idea seems interesting to me, but I don't quite know how to implement it unless I capacitor couple the input transformer secondary to the transistor bases.
That would add two caps and two resistors and reduce gain somewhat, but it would also be a bit of feedback which might make the amp sound better.
Oh I see so that's 42Vrms maximum across the full 600 ohm primary, right? If so that would be 84Vrms out of the secondary given the transformer has a 1:2 voltage ratio. Does having the bias set where it will be reduce gain? If so I will see if I can get to 50Vrms out of the transformer for the same input voltage. If I cannot then would a 2N2222A work to provide enough gain?
I will need to ask Edcor just what the 50Vrms voltage limit is for the 600 ohm to 2.4K transformer as I don't know if that is just the limitation of one winding or the maximum voltage that can be applied to or output of either winding.
If the 50Vrms limit is just for the 600 ohm winding then I can indeed have much more output voltage using the diodes and a different bias point. That would also reduce the chance of instability and I may then not need the .0033uF cap.
When I get the amp finished I will be using a LM-317 set for 24Vdc to provide the B+ voltage. The MK-484 kit I have ordered has its own voltage divider as it also has a chip amp on it that does at least a watt of power and the circuit can run at up to 9Vdc I think so I could use a 7809 after the 24 volt regulator to provide its voltage. I could have just used that and an 8 ohm to 5K transformer to properly drive the speaker, but I don't think that's necessarily the best way to go.
Transistors don't like inductive source impedance. Try a resistor in series with the base, 1k +/-.
Maybe bias was too low. You need enough collector current for r_e to become <<10 ohms. That's >>2.6 mA.
Whatever floats your boat. If resulting bias is too low, try something 1N4148-ish, if it's too high, go for 1N54xx instead.
No. Useless for thermal tracking.
Which brings us to where the bias diodes go. Two of them should be thermally coupled to one of the Darlingtons. (I think it'll be overcorrected with all three.) Soldering directly to a transistor leg may prove a fair bit quicker than heat transfer over plastic, but YMMV.
> one side is slightly lower than the other
Yes, I realized that you have NO scheme to force balance between the two transistors. They never match exactly, and this is a HIGH gain circuit, so small difference of Vbe between transistors will be a large difference in idle current.
Put 1 Ohm in series with each emitter. This will drop 35mV-50mV, which is probably larger than the Vbe difference of two Darlingtons from the same lot. Check that the voltage drop is similar over each 1 Ohm resistor, say within 20%.
Do you have any heatsink on the transistors? At 1.2 Watts each you are running very hot for a naked TO-66. Obviously if you put it on a pound of alloy in a cold lake there would be "no" temperature rise. A couple square inches of fairly thick flat Aluminum may be a good size.
The diodes should be on the heatsink. Diode tracking is inexact and I would not be too concerned about how many are on the loop.
If your power output is ample and your real problem is current stability for gain stability, throw-away much of your supply voltage in a large emitter resistor. While a regulator "aint right", if we hold the Bases up at say +5V then the Vbe variation is only 25% of the total bias. Common emitter resistor could be 50 or 55 Ohms to get ~~70mA idle current.
> what the 50Vrms voltage limit
The basic insulation is 500V. A "50V" limit is for undistorted bass at a specified frequency. They casually drop numbers of "20Hz" and "0.05%". I doubt both apply at full level, but I'm sure your useful low-limit is at least an octave higher. Max voltage for specified THD scales roughly with frequency. Whatever THD it has at say 50V 50Hz, it will be the same at 100hz and _100V_.
Yes, I realized that you have NO scheme to force balance between the two transistors. They never match exactly, and this is a HIGH gain circuit, so small difference of Vbe between transistors will be a large difference in idle current.
Put 1 Ohm in series with each emitter. This will drop 35mV-50mV, which is probably larger than the Vbe difference of two Darlingtons from the same lot. Check that the voltage drop is similar over each 1 Ohm resistor, say within 20%.
Do you have any heatsink on the transistors? At 1.2 Watts each you are running very hot for a naked TO-66. Obviously if you put it on a pound of alloy in a cold lake there would be "no" temperature rise. A couple square inches of fairly thick flat Aluminum may be a good size.
The diodes should be on the heatsink. Diode tracking is inexact and I would not be too concerned about how many are on the loop.
If your power output is ample and your real problem is current stability for gain stability, throw-away much of your supply voltage in a large emitter resistor. While a regulator "aint right", if we hold the Bases up at say +5V then the Vbe variation is only 25% of the total bias. Common emitter resistor could be 50 or 55 Ohms to get ~~70mA idle current.
> what the 50Vrms voltage limit
The basic insulation is 500V. A "50V" limit is for undistorted bass at a specified frequency. They casually drop numbers of "20Hz" and "0.05%". I doubt both apply at full level, but I'm sure your useful low-limit is at least an octave higher. Max voltage for specified THD scales roughly with frequency. Whatever THD it has at say 50V 50Hz, it will be the same at 100hz and _100V_.
I did the diode mod while leaving the variable resistor in place and found that in order to get 75mA B+ current at idle I needed the resistor set to 4.49K.
Bias voltage is 1.867Vdc at the center tap of the input transformer secondary.
Bias voltage to one transistor base is 1.863Vdc.
Bias voltage to the other transistor base is 1.861Vdc.
Bias current at 50Vrms output at 800Hz is 75.35mA.
So there is a small difference due to winding resistance, but doesn't look like enough to matter.
Now when I get the MK-484 kit if I find gain is not enough I can just use 1/2 the primary of the input transformer and I will then have a 1:2 voltage ratio.
Set the amp for 50Vrms output and it settled at 50.3Vrms. Turned it off and let it cool to room temperature. Upon turning on the amp the output voltage started at 49.6Vrms so there is a total of 600mV of variance between initial turn on and when the amp has been running for awhile. That amount of difference is ok with me.
I did take the diodes and mount them to where they touch the chassis between both transistors.
For 50Vrms output using the whole primary winding I need an input of 69.34mVrms.
For 50Vrms output using 1/2 the primary winding I need an input of 35.09mVrms.
What I might do is replace the resistor supplying the voltage to the MK-484 with an audio taper pot of the same value and just use that as the volume control with it capacitor coupled to the input transformer unless that wouldn't be a good idea.
I now have a good clean sinewave down to 35Hz and the output voltage has dropped to 44.38Vrms (did drop more until I realized the input voltage dropped as well) which could be due to the transformer itself or the fact that the primary of the output transformer has dc current going through it when from what I understand it isn't designed for.
Also at 35Hz if I take the output voltage above 50Vrms on the secondary the sinewave starts to distort.
grossklass,
Trying 1K in series with the bases would affect bias, right?
Originally I had the diodes in series with the variable resistor and not from the center tap of the secondary to ground.
The diodes are touching the chassis between where the two transistors are mounted to the chassis and that arrangement seems to work quite well.
tauro0221,
The microcontroller idea is interesting, but is a bit overkill for what I am doing.
PRR,
I will have to try the 1 ohm resistors.
The transistors are mounted to the metal chassis I used. The transistors themselves aren't getting above 80F.
The amp itself is real stable now far as gain is concerned.
75mA seems to be the perfect current for the circuit.
When I got the amp home and tried it on the Peerless reproducer I did have a bit of feedback which the audible bit was solved by connecting the input ground to the main ground.
I also had to readjust bias some to get the 75mA and the audio isn't quite as clear so I may have another oscillation that I cannot hear. Will try a cap across the bases tomorrow.
Not sure why the feedback now when it worked so good with the 5K load, but I suspect it is because the driver depending on frequency can have an impedance of a little above 10K. One reason I know I may have inaudible feedback is because without any audio being fed to the amp, the volume control affects bias slightly even though it is isolated from the circuit by a transformer.
I will try increasing the value of the 5K variable resistor used as a load until I start to get feedback then add a cap across the input transformer secondary and adjust its value until the feedback is gone and the treble isn't reduced too much. As long as the amp is relatively flat to 5KHz that will be good enough for my purposes.
Bias voltage is 1.867Vdc at the center tap of the input transformer secondary.
Bias voltage to one transistor base is 1.863Vdc.
Bias voltage to the other transistor base is 1.861Vdc.
Bias current at 50Vrms output at 800Hz is 75.35mA.
So there is a small difference due to winding resistance, but doesn't look like enough to matter.
Now when I get the MK-484 kit if I find gain is not enough I can just use 1/2 the primary of the input transformer and I will then have a 1:2 voltage ratio.
Set the amp for 50Vrms output and it settled at 50.3Vrms. Turned it off and let it cool to room temperature. Upon turning on the amp the output voltage started at 49.6Vrms so there is a total of 600mV of variance between initial turn on and when the amp has been running for awhile. That amount of difference is ok with me.
I did take the diodes and mount them to where they touch the chassis between both transistors.
For 50Vrms output using the whole primary winding I need an input of 69.34mVrms.
For 50Vrms output using 1/2 the primary winding I need an input of 35.09mVrms.
What I might do is replace the resistor supplying the voltage to the MK-484 with an audio taper pot of the same value and just use that as the volume control with it capacitor coupled to the input transformer unless that wouldn't be a good idea.
I now have a good clean sinewave down to 35Hz and the output voltage has dropped to 44.38Vrms (did drop more until I realized the input voltage dropped as well) which could be due to the transformer itself or the fact that the primary of the output transformer has dc current going through it when from what I understand it isn't designed for.
Also at 35Hz if I take the output voltage above 50Vrms on the secondary the sinewave starts to distort.
grossklass,
Trying 1K in series with the bases would affect bias, right?
Originally I had the diodes in series with the variable resistor and not from the center tap of the secondary to ground.
The diodes are touching the chassis between where the two transistors are mounted to the chassis and that arrangement seems to work quite well.
tauro0221,
The microcontroller idea is interesting, but is a bit overkill for what I am doing.
PRR,
I will have to try the 1 ohm resistors.
The transistors are mounted to the metal chassis I used. The transistors themselves aren't getting above 80F.
The amp itself is real stable now far as gain is concerned.
75mA seems to be the perfect current for the circuit.
When I got the amp home and tried it on the Peerless reproducer I did have a bit of feedback which the audible bit was solved by connecting the input ground to the main ground.
I also had to readjust bias some to get the 75mA and the audio isn't quite as clear so I may have another oscillation that I cannot hear. Will try a cap across the bases tomorrow.
Not sure why the feedback now when it worked so good with the 5K load, but I suspect it is because the driver depending on frequency can have an impedance of a little above 10K. One reason I know I may have inaudible feedback is because without any audio being fed to the amp, the volume control affects bias slightly even though it is isolated from the circuit by a transformer.
I will try increasing the value of the 5K variable resistor used as a load until I start to get feedback then add a cap across the input transformer secondary and adjust its value until the feedback is gone and the treble isn't reduced too much. As long as the amp is relatively flat to 5KHz that will be good enough for my purposes.
Last edited:
Somewhat, depending on bias current flowing there. Maybe even a 330-470 ohm will do to eliminate the issue, but you'll have to experiment.
Problem solved.
Here's what I did to fix it.
1. All wiring carrying audio replaced with shielded cable.
2. Removed cap on output transformer primary as it made the oscillation worse.
3. Added a .0033uF cap to ground from each transistor base.
Now the amp can be completely unloaded and not oscillate at all and I can go up to 60Vrms output unloaded before any signs of oscillation.
There is a problem I've noticed.
I've been testing the amp at 800 Hz and 8KHz to see what effect the capacitor had on the signal and it doesn't attenuate the signal, but what does happen is this. As the load impedance increases the difference in output voltage between 800Hz and 8KHz gets bigger.
At 5K load there is only 5Vrms difference, but with a 20K load the difference is 15Vrms.
I Am thinking it could be due to the impedance being much higher than the 2.5K the transformer was designed for.
Now to test the amp more at work and see just how stable it is then the real test is when I get it home and test it with the speaker. When I get home from work I can post an impedance graph of the speaker to show y'all what I'm having to work with.
Here's what I did to fix it.
1. All wiring carrying audio replaced with shielded cable.
2. Removed cap on output transformer primary as it made the oscillation worse.
3. Added a .0033uF cap to ground from each transistor base.
Now the amp can be completely unloaded and not oscillate at all and I can go up to 60Vrms output unloaded before any signs of oscillation.
There is a problem I've noticed.
I've been testing the amp at 800 Hz and 8KHz to see what effect the capacitor had on the signal and it doesn't attenuate the signal, but what does happen is this. As the load impedance increases the difference in output voltage between 800Hz and 8KHz gets bigger.
At 5K load there is only 5Vrms difference, but with a 20K load the difference is 15Vrms.
I Am thinking it could be due to the impedance being much higher than the 2.5K the transformer was designed for.
Now to test the amp more at work and see just how stable it is then the real test is when I get it home and test it with the speaker. When I get home from work I can post an impedance graph of the speaker to show y'all what I'm having to work with.
Last edited:
Please post your modified schematic.
As shown in the cirst one, that´s not the way to connect a Zener and IF you replaced it with some diodes, the error persists.
It´s "regulating" nothing, in fact it *increases* ripple and error because all it does is apply the full voltage availble on the bias string, which is NOT regulatednor thermally stabilized, *shifter* by a fixed amount, in the case of the Zenerfixed 15V, so proportionally errrors annd jun become *lrger* than if you just used all resistors, go figure.
But I´m guessing based on a later circuit you didn´t post so please do.
Thanks.
As shown in the cirst one, that´s not the way to connect a Zener and IF you replaced it with some diodes, the error persists.
It´s "regulating" nothing, in fact it *increases* ripple and error because all it does is apply the full voltage availble on the bias string, which is NOT regulatednor thermally stabilized, *shifter* by a fixed amount, in the case of the Zenerfixed 15V, so proportionally errrors annd jun become *lrger* than if you just used all resistors, go figure.
But I´m guessing based on a later circuit you didn´t post so please do.
Thanks.
Here's the updated schematic as things are currently with the amp.
I got the amp home and I discovered an interesting problem.
With the speaker connected and the volume control either full CCW or full CW I get an oscillation. I have no real clue what is causing it unless the capacitor that is across the input of the speaker is the problem.
I tried a 7K variable in series with the speaker and transformer and adjusted it until the oscillation stopped. Came to 720 something ohms so I used an 820 ohm resistor.
I did notice that the voltage across the resistor does increase somewhat as the frequency goes lower which is expected given the impedance of the driver goes lower as the frequency goes lower.
Also even though the sinewave looks clean at 800Hz up to a little above 50Vrms if I play music loud enough until I hear distortion the average reading is around 35Vrms. Suppose that is to be expected of any amplifier given the amp is not just amplifying a single frequency, but may be exaggerated more with the Peerless speaker due to how much the impedance of it varies.
Impedance graph of the speaker.
I did try adding the 1 ohm resistors and re-adjusting bias for 75mA, but I don't know if that really helped much.
I got the amp home and I discovered an interesting problem.
With the speaker connected and the volume control either full CCW or full CW I get an oscillation. I have no real clue what is causing it unless the capacitor that is across the input of the speaker is the problem.
I tried a 7K variable in series with the speaker and transformer and adjusted it until the oscillation stopped. Came to 720 something ohms so I used an 820 ohm resistor.
I did notice that the voltage across the resistor does increase somewhat as the frequency goes lower which is expected given the impedance of the driver goes lower as the frequency goes lower.
Also even though the sinewave looks clean at 800Hz up to a little above 50Vrms if I play music loud enough until I hear distortion the average reading is around 35Vrms. Suppose that is to be expected of any amplifier given the amp is not just amplifying a single frequency, but may be exaggerated more with the Peerless speaker due to how much the impedance of it varies.
Impedance graph of the speaker.
I did try adding the 1 ohm resistors and re-adjusting bias for 75mA, but I don't know if that really helped much.
Attachments
Last edited:
Took the speaker and amp to work to figure out what was causing the problem.
Turns out it was the capacitor in the speaker. Must have been forming a resonant circuit with the transformer secondary.
The speaker is wired like this.
Cap across input wires. Cap across driver coil. Choke connecting the two.
I removed the cap across the input wires and the amp is good now.
I also fed some music to the amp and the speaker plays nicely.
Did that to be sure the amp indeed did work properly.
Now if it doesn't work right at home I will know it is not the amp or speaker.
The power supply I use at home is a regulated 12 volt Astron that was used to power Motorola radios, but I use the 24Vdc across the main filter cap which is unregulated. Could that be causing an issue using the amp at home as I use a regulated supply at work?
Just now thought of this thing.
I could get a 30Vdc AGM or led acid battery (whichever standard battery is closest to that voltage) and have a portable AM radio that can receive my part 15 AM broadcaster for doing things like yard work and given the circuit only draws 75mA plus whatever the MK-484 circuit draws it should be able to run a very long time on a battery. I would then build a charging circuit that will keep the battery charged and power the circuit. I could even get a solar cell to keep the battery topped off or possibly dispense with the battery altogether and just use a solar cell. The charging circuit would be such that no more than the proper charging current is fed to the battery including what the circuit draws unless it is better to not have the circuit connected while the battery is charging and if so I can use a relay to switch the battery to the charging circuit and switch the amp circuitry to the power supply operating the charging circuit.
Would make it to where I am not limited to being near a power outlet.
Also would be great to have during a power outage.
Turns out it was the capacitor in the speaker. Must have been forming a resonant circuit with the transformer secondary.
The speaker is wired like this.
Cap across input wires. Cap across driver coil. Choke connecting the two.
I removed the cap across the input wires and the amp is good now.
I also fed some music to the amp and the speaker plays nicely.
Did that to be sure the amp indeed did work properly.
Now if it doesn't work right at home I will know it is not the amp or speaker.
The power supply I use at home is a regulated 12 volt Astron that was used to power Motorola radios, but I use the 24Vdc across the main filter cap which is unregulated. Could that be causing an issue using the amp at home as I use a regulated supply at work?
Just now thought of this thing.
I could get a 30Vdc AGM or led acid battery (whichever standard battery is closest to that voltage) and have a portable AM radio that can receive my part 15 AM broadcaster for doing things like yard work and given the circuit only draws 75mA plus whatever the MK-484 circuit draws it should be able to run a very long time on a battery. I would then build a charging circuit that will keep the battery charged and power the circuit. I could even get a solar cell to keep the battery topped off or possibly dispense with the battery altogether and just use a solar cell. The charging circuit would be such that no more than the proper charging current is fed to the battery including what the circuit draws unless it is better to not have the circuit connected while the battery is charging and if so I can use a relay to switch the battery to the charging circuit and switch the amp circuitry to the power supply operating the charging circuit.
Would make it to where I am not limited to being near a power outlet.
Also would be great to have during a power outage.
30 Volts of Lead is barely "portable".
Wasn't the battery transistor radio invented already? OK, few have a 1+ Watt output and a vintage speaker.
The solar battery will be many-many cells (0.6V per cell), and a reasonable price array will just about work in bright sunlight.
For extended and dark power outages, a "9V" pocket radio will run for months on a used 6V lantern battery or for some weeks on a charged 4-cell alarm battery.
Wasn't the battery transistor radio invented already? OK, few have a 1+ Watt output and a vintage speaker.
The solar battery will be many-many cells (0.6V per cell), and a reasonable price array will just about work in bright sunlight.
For extended and dark power outages, a "9V" pocket radio will run for months on a used 6V lantern battery or for some weeks on a charged 4-cell alarm battery.
Thanks to all who have helped with the amp.
It works great.
Seems like speaker impedance doesn't affect the amp as much as I thought it would when play
Just tried it with an AK E3 speaker and realized the speaker needs work so I tried it with a Radiola 100 speaker and it works great.
The amp does drive the speakers to a quite loud listening level. I initially thought the amp was putting out way more voltage than what would be enough to damage an antique high impedance speaker, but it does not put out too much voltage.
I do plan on using this with the MK-484, but I also may build another one just for testing antique speakers.
I have found that the antique speakers sound a bit better when fed from this amp which could be due to the fact that a lot of the 1920's radios didn't have much bass response to begin with.
I like how simple this amp is.
It works great.
Seems like speaker impedance doesn't affect the amp as much as I thought it would when play
Just tried it with an AK E3 speaker and realized the speaker needs work so I tried it with a Radiola 100 speaker and it works great.
The amp does drive the speakers to a quite loud listening level. I initially thought the amp was putting out way more voltage than what would be enough to damage an antique high impedance speaker, but it does not put out too much voltage.
I do plan on using this with the MK-484, but I also may build another one just for testing antique speakers.
I have found that the antique speakers sound a bit better when fed from this amp which could be due to the fact that a lot of the 1920's radios didn't have much bass response to begin with.
I like how simple this amp is.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.