I've had an ongoing project for an amplifier for my main system, which has resulted in a 135W stereo amplifier. For a few reasons I'm now using it for a secondary system, and the project is still on. For a few years before this amplifier, I made good use of a 35W stereo amplifier. This amplifier has been decommissioned, so I have a spare transformer which needs an amplifier.
I decided to use it for a bridgeable amplifier, so there will be 4 amplifiers. I've subsequently decided to rather use it to bi-amp, though, but that's another subject. I'm going to post a few pics and some info. Enjoy!
The design is a simple voltage F/B design. I've experimented with a symmetrical design, and was extremely happy with the result (incredible quality, and absolutely no turn on or off noises whatsoever). But I'm not done with my experimenting with it, and I also needed an amplifier able to drive 4 ohms. I need to do more experimenting with the symmetrical amplifier's class A stage, because it does draw quite a lot of current, and is quite susceptible to thermal runaway.
I'm using 2SA1837/2SC4793 for the drivers and 2SA1186/2SC2837 for the output. Small signal transistors are 2N5551. The preamplifier is a very basic class A voltage F/B design and a unity gain inverting op amp circuit to provide the inverted signal for two of the amplifiers. The pnp small signal transistors are 2N5401. The op amp is an NE5532.
At the moment I'm missing my power transistors and some capacitors. Hopefully they arrive today.
Anyway, here is the PCB so far:
I decided to use it for a bridgeable amplifier, so there will be 4 amplifiers. I've subsequently decided to rather use it to bi-amp, though, but that's another subject. I'm going to post a few pics and some info. Enjoy!
The design is a simple voltage F/B design. I've experimented with a symmetrical design, and was extremely happy with the result (incredible quality, and absolutely no turn on or off noises whatsoever). But I'm not done with my experimenting with it, and I also needed an amplifier able to drive 4 ohms. I need to do more experimenting with the symmetrical amplifier's class A stage, because it does draw quite a lot of current, and is quite susceptible to thermal runaway.
I'm using 2SA1837/2SC4793 for the drivers and 2SA1186/2SC2837 for the output. Small signal transistors are 2N5551. The preamplifier is a very basic class A voltage F/B design and a unity gain inverting op amp circuit to provide the inverted signal for two of the amplifiers. The pnp small signal transistors are 2N5401. The op amp is an NE5532.
At the moment I'm missing my power transistors and some capacitors. Hopefully they arrive today.
Anyway, here is the PCB so far:
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I have a few more pics I would like to share. My transistors arrived.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Each amplifier on the board has 2 diodes, these have now got a blob of heat paste on, and are bent onto a driver transistor each. The class A transistor has a heatsink on now as well (operates at 300 mW). The board is missing 2 capacitors for the preamplifier's power supply. I'll be finishing the amplifier this weekend, and I'll load it into an enclosure in the next few weeks.
I've used 8 3300 uF capacitors per rail, giving very good ripple current capability (about 18 A per rail). It's also so cheap to do this. In the past I've used 10000 uF capacitors, but they're expensive, and can't compete with the specification of all these 3300 uF's.
This amplifier will easily drive 4 ohm loads, and will do about 65 W into 4 ohms with the power supply it was designed for. Changing one resistor will allow it to use up to 43-0-43 Vdc rails for 100 W into 8 ohms without changing a single extra thing (except some components in the power supply, obviously).
The transistors don't look neat; this is because I twisted them slightly here and there so that their mounting holes would match up with a gap in the heatsink fins. Nothing wrong with a little untidiness if the benefit is large.
I'll maybe post the schematic sometime.
This amplifier has no trimmer pot to set the bias. It's designed to run with a bias current of 300 mA (this was a painful exercise). This will give a few watts of class A, which is nice, and there should be no crossover distortion. I usually design for 25 - 30 mA, so 300 mA is a bit out of my comfort zone. Each amplifier will dissipate between 15 and 20 W quiescent, so the heatsinks will run at around 50 - 55 degrees C at an ambient of 30 (heatsinks are 0.7 K/W).
I've used 8 3300 uF capacitors per rail, giving very good ripple current capability (about 18 A per rail). It's also so cheap to do this. In the past I've used 10000 uF capacitors, but they're expensive, and can't compete with the specification of all these 3300 uF's.
This amplifier will easily drive 4 ohm loads, and will do about 65 W into 4 ohms with the power supply it was designed for. Changing one resistor will allow it to use up to 43-0-43 Vdc rails for 100 W into 8 ohms without changing a single extra thing (except some components in the power supply, obviously).
The transistors don't look neat; this is because I twisted them slightly here and there so that their mounting holes would match up with a gap in the heatsink fins. Nothing wrong with a little untidiness if the benefit is large.
I'll maybe post the schematic sometime.
This amplifier has no trimmer pot to set the bias. It's designed to run with a bias current of 300 mA (this was a painful exercise). This will give a few watts of class A, which is nice, and there should be no crossover distortion. I usually design for 25 - 30 mA, so 300 mA is a bit out of my comfort zone. Each amplifier will dissipate between 15 and 20 W quiescent, so the heatsinks will run at around 50 - 55 degrees C at an ambient of 30 (heatsinks are 0.7 K/W).
I built up the courage to turn this one on last night. First thing, big spark. On inspection, I found my secondary power supply had two diodes soldered in the wrong way, so a thin track burned. Fixed it up, turned it on, everything seemed to work 100% as designed - 100mV over one of the 0.33R output resistors (giving 309 mA quiescent). I haven't done more testing than that because it seems one of my output transistors is faulty. There has been no fault in the amplifier, and three of them work as expected. One of the transistors measures like a capacitor on the diode test. I will replace it and probably surrounding transistors tonight and see what happens. I found this issue because one of my driver resistors started to smoke (the one on the faulty transistor).
I've never experienced something like this, and I may swap my output transistors to my old faithful and super-reliable TIP35C and TIP36Cs. I'm not sure if the transistors I got are genuine or not. I'll see how it goes.
I've never experienced something like this, and I may swap my output transistors to my old faithful and super-reliable TIP35C and TIP36Cs. I'm not sure if the transistors I got are genuine or not. I'll see how it goes.
Transistors are fake! What a disappointment! The die is 2mm x 2mm. It's bonded directly onto the tab and the bonding wires are microscopic. They fail completely randomly at 26V and 300mA. And don't fail like normal. I'll replace with tip35 or tip41. They have worked great for me in the past, especially tip41's.
Here is one of the transistors opened up (that's the die there):
An externally hosted image should be here but it was not working when we last tested it.
I calculate a dissipation from one channel @ 25.8W (0.3A*(43+43V))
Two channels into a 0.7C/W sink gives a delta T ~36Cdegrees and apply a factor of 1.1 to 1.2 since delta T <<70Cdegrees and I get Ts~73°C
Tc will be around 80°C without any audio signal passing.
Applying a de-rating factor of 1.3 gives Ts~58°C and Tc~65°C
Why did you select 100mV for Vre?
That is a massive increase above optimal ClassAB, where crossover distortion is increased for a ClassA output of <0.7W into 4r0
Thanks for the question.
I must look into that optimal classAB you mention.
My idea with this amplifier is no-setup with predictable bias points. The 100 mV is convenient because 1N4007 diodes have 100 mV more diode drop than my driver transistors, giving the 100 mV on the output resistors. The diodes and transistors also behave similarly at different temperatures (although the amplifier is designed that the drivers shouldn't heat up more than a few degrees above ambient). With a relatively small output resistor of 0R33, I have a reasonable quiescent current of 300 mA, so all in all it's a compromise, and a convenient one at that. With those output resistors, I'm sacrificing a mere 1 or 2 W (8 ohms load).
Furthermore, it makes construction much easier, since the power supply is on-board, so I don't have to fiddle to measure current. I need to only measure 1 amplifier (obviously each must be checked), and I can measure for an exact design-point. This is especially nice when building 1 board with 4 amplifiers.
Oh, and I did look into many diodes. The 1N400x were definitely the best option all round.
As a matter of interest, my output stage should have no visible crossover distortion without any global feedback. It's a simple, but effective output stage, very similar to what Elliott uses in his 3A.
I must look into that optimal classAB you mention.
My idea with this amplifier is no-setup with predictable bias points. The 100 mV is convenient because 1N4007 diodes have 100 mV more diode drop than my driver transistors, giving the 100 mV on the output resistors. The diodes and transistors also behave similarly at different temperatures (although the amplifier is designed that the drivers shouldn't heat up more than a few degrees above ambient). With a relatively small output resistor of 0R33, I have a reasonable quiescent current of 300 mA, so all in all it's a compromise, and a convenient one at that. With those output resistors, I'm sacrificing a mere 1 or 2 W (8 ohms load).
Furthermore, it makes construction much easier, since the power supply is on-board, so I don't have to fiddle to measure current. I need to only measure 1 amplifier (obviously each must be checked), and I can measure for an exact design-point. This is especially nice when building 1 board with 4 amplifiers.
Oh, and I did look into many diodes. The 1N400x were definitely the best option all round.
As a matter of interest, my output stage should have no visible crossover distortion without any global feedback. It's a simple, but effective output stage, very similar to what Elliott uses in his 3A.
Thanks, I'll look into it. I built up my knowledge from applying my knowledge I learned at university, and a book called The Art of Electronics. I have an affinity for BJTs. I've read many resources online, but I have a huge amount to learn. I understand transistor circuits very well, but I want to know more, so thank you for the referral.
I want to build an output like this:
An externally hosted image should be here but it was not working when we last tested it.
It has great rail-to-rail performance, so I just want to play around with it. I have a symmetrical amplifier which has fantastic performance, but bad thermal stability, and I want to explore the design further. It will be my next project. I'm thinking of using the above output with it, because the symmetrical design already takes away from possible maximum swing, so they should go well together.
Oscillations. Goodness, have I struggled with those! Thankfully I have found a few ways to mitigate it, one being low gain. This 4 channel amplifier has very low gain - the preamplifier has about 3 V/V (9.5 dB) gain and the amplifiers have about 12 V/V (21.6 dB).
So I've bought TIP41C / TIP42C to replace the fake SanKen's. I hope the Toshiba drivers I got are at least real. I've had great results from TIP41/2 in the past, so I'm looking forward to getting these in!
Once in, I'll do a few tests, and if all's well, I'll make some measurements and post the results. I can't measure distortion, but I'm not a distortion freak. I can't hear 0.1% distortion vs 0.0001% distortion, even on a good pair of B&W's. And this amplifier simulates to 0.01%, and I'm fine with that.
Once in, I'll do a few tests, and if all's well, I'll make some measurements and post the results. I can't measure distortion, but I'm not a distortion freak. I can't hear 0.1% distortion vs 0.0001% distortion, even on a good pair of B&W's. And this amplifier simulates to 0.01%, and I'm fine with that.
I've done a lot of thinking and simulating, and the amplifier should reject hum well. The real-life measurements are similar to simulation to within 5% or so. The hum is far more than would be expected from power supply ripple making it to the outputs. It also seems that the hum comes largely from the preamplifier, since it is much more prominent on the LF amplifiers. I believe I have a few grounding issues, but I think most of all, I need to do some checking on the preamplifier (also grounding there). It's going to take a while, but I'm confident I can remove all hum to below-audible. I will post my results of this so that it may be useful to others.
First thing I'm going to do is replace my long-tail resistors with proper current sources. This should be fairly easy, and give much better power supply rejection (by at least two orders of magnitude).
This should also reduce THD significantly.
(Lesson learned)
An externally hosted image should be here but it was not working when we last tested it.
This should also reduce THD significantly.
(Lesson learned)
Last edited:
Short the Power Amplifier inputs with zero ohm dummy phono plug loads.
Measure the output noise and offset using the 199.9mVac & dc scales of your DMM.
Then add on two interconnects and short the remote ends and measure again.
Then short the barrels, at the remote ends, together and measure again.
This last test often tells whether your amplifier wiring is good, or bad.
You do not need and should not use a pre-amplifier to test your Power Amplifier.
Thanks for the tips. I have done pretty much that to no effect. Everything remains the same.
I changed the way I grounded this amplifier by not gounding my shielding on both sides for signal paths. I also connected the power supply differently. There are a number of things I know can have a big impact, and I'll experiment with those and post the results. I'm pretty sure my wiring isn't great.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.