Member
Joined 2009
Paid Member
http://www.diyaudio.com/forums/solid-state/133189-my-first-diy-amplifier-20-years-go.html
for some reason I thought you knew of this amplifier, but I see you never posted into that thread so maybe not. It's a symmetric amplifier with BJTs that reminds me of the amp in this thread although they are not the same.
for some reason I thought you knew of this amplifier, but I see you never posted into that thread so maybe not. It's a symmetric amplifier with BJTs that reminds me of the amp in this thread although they are not the same.
Unfortunately I have not heard this amp. The output stage is an emitter follower BJT so it is quite different from the amp in this thread. For very low power the emitter follower stage may not be needed...
Fab
Little known fact:
With 0.47 ohms and especially 0.33 ohms you still enjoy a certain amount of
"square law" action, so the 1 A bias will actually typically give you more Class A
output than that. Probably less that 16 watts into 4.
Interesting the square law effect for lower impedance load...
Thanks for this information
Fab
Bumping an oldish thread:
To increase gain and run off higher rails, should we increase the feedback resistor or decrease the Emitter resistors?
I'm building three of these to power the front channels of a small gaming/surround setup as this is the simplest possible way to use some of the dozen pairs of IRFP parts and spare trimmers/thermistors I have laying around from a bunch of F5 and similar builds, and without needing to import unobtanium at jacked-up prices or definite fakes from China.
I was also considering the 'Mullard style amps from LXG, they sound brilliant but I'm unable to convince myself that a simple power supply can handle charging the bias caps fast enough.
I'm planning to put these on a plate amp, so obviously the bias will be way lower than the stock 1.3A, maybe only about 750mA for the fronts and half that for the center. Maybe even lower depending on the plate design. Since this is late-night gaming, high volumes are not necessary. Rest of the channels will run off chipamp type solutions.
Anywho, I need the amp to swing ~20V into 8 ohm load (Dayton RS100-8s) from a 1.3Vrms source, so a gain of about 20-22dB is what I might need. I will be running ~30V rails from a 240VA transformer.
I could achieve this by lowering degen or feedback, but which might be the preferred method? I assume the feedback is easier to do and will not muck up the front end bias too much? Thinking about this the right way?
To increase gain and run off higher rails, should we increase the feedback resistor or decrease the Emitter resistors?
I'm building three of these to power the front channels of a small gaming/surround setup as this is the simplest possible way to use some of the dozen pairs of IRFP parts and spare trimmers/thermistors I have laying around from a bunch of F5 and similar builds, and without needing to import unobtanium at jacked-up prices or definite fakes from China.
I was also considering the 'Mullard style amps from LXG, they sound brilliant but I'm unable to convince myself that a simple power supply can handle charging the bias caps fast enough.
I'm planning to put these on a plate amp, so obviously the bias will be way lower than the stock 1.3A, maybe only about 750mA for the fronts and half that for the center. Maybe even lower depending on the plate design. Since this is late-night gaming, high volumes are not necessary. Rest of the channels will run off chipamp type solutions.
Anywho, I need the amp to swing ~20V into 8 ohm load (Dayton RS100-8s) from a 1.3Vrms source, so a gain of about 20-22dB is what I might need. I will be running ~30V rails from a 240VA transformer.
I could achieve this by lowering degen or feedback, but which might be the preferred method? I assume the feedback is easier to do and will not muck up the front end bias too much? Thinking about this the right way?
Thanks for the input Andrew, that's what I thought intuitively.
The power dissipation in the feedback resistor goes from about a watt to just under 1.5W when I increase the rails to 30V from 24.
Which is where the thought about decreasing the degeneration came from in the first place, though I will admit I hadn't crunched the numbers. I was looking at the typical range of values for the JFET version and wondering if I should look at lowering the 150 ohm resistors to say ~47-68 ohms. The increase in FE bias is minimal so the BC5xx should be fine.
I may have been overthinking the whole thing, but lower value resistors seem to create less noise etc, hence the question.
The power dissipation in the feedback resistor goes from about a watt to just under 1.5W when I increase the rails to 30V from 24.
Which is where the thought about decreasing the degeneration came from in the first place, though I will admit I hadn't crunched the numbers. I was looking at the typical range of values for the JFET version and wondering if I should look at lowering the 150 ohm resistors to say ~47-68 ohms. The increase in FE bias is minimal so the BC5xx should be fine.
I may have been overthinking the whole thing, but lower value resistors seem to create less noise etc, hence the question.
Thanks for the help to Andrew and the great circuit to juma. I have three of these mounted on a plate and have the fronts biased to about 400mA and the center to about 250, all sound great. I'm currently testing without thermistors so I can work out what level of bias the heatsink is comfortable with.
In order to support the higher gain (4k7/249E because I had these on hand) I used two BC550/560 per board as the amp was quite clearly running out of steam at high output level, with lots of HF distortion. The situation improved somewhat with two pairs but I suspect four might be better. My audio card is OOO so can't get THD measurements for a bit yet, but sometime soon.
The biggest issue was the use of the flat single-turn pots. I should really advise against using these as the adjustment process becomes extremely tricky and getting offset below 10mV or so is practically impossible.
In order to support the higher gain (4k7/249E because I had these on hand) I used two BC550/560 per board as the amp was quite clearly running out of steam at high output level, with lots of HF distortion. The situation improved somewhat with two pairs but I suspect four might be better. My audio card is OOO so can't get THD measurements for a bit yet, but sometime soon.
The biggest issue was the use of the flat single-turn pots. I should really advise against using these as the adjustment process becomes extremely tricky and getting offset below 10mV or so is practically impossible.
That brought the trouble...
To improve things, let feedback resistors be 150R/22R, make feedback caps=4700uF. Depending on the rail voltage adjust the value of emitter-feeding resistors so that that you get about 4-5mA through BJTs and then one pair of input BJT will be enough and the amp will work and sound as it was supposed to.
Hi Juma
I thought as much, and was planning to back down the resistors before closing the plate, changing them when I added the thermistors and did final setup. Thanks for the help on the values.
I reduced the emitter resistors in proportion to the rail voltage so they're 2.7k for 19V rails. Hope that is good. I'm running these off laptop bricks in series, no noise on output (the bricks are decent current-mode units with very low noise).
Would you be able to tell me what gain I can shoot for using just one pair? I need to get enough gain to get max output from low-output devices like phones and PC soundcards (not the professional ones I normally use, these ones have 300-500mV outputs). I'm guessing 20-24dB is what I will definitely need.
I thought as much, and was planning to back down the resistors before closing the plate, changing them when I added the thermistors and did final setup. Thanks for the help on the values.
I reduced the emitter resistors in proportion to the rail voltage so they're 2.7k for 19V rails. Hope that is good. I'm running these off laptop bricks in series, no noise on output (the bricks are decent current-mode units with very low noise).
Would you be able to tell me what gain I can shoot for using just one pair? I need to get enough gain to get max output from low-output devices like phones and PC soundcards (not the professional ones I normally use, these ones have 300-500mV outputs). I'm guessing 20-24dB is what I will definitely need.
What you need is one thing and what you can get is quite another. If you want it to sound right don't wander far from the values I mentioned in the previous post.
Have fun !
Got it juma, thanks. I shall take it from here on out. I shall keep maybe 165 and 19.5 (paralleled combos of 330 and 39 ohms) that will get me into the ballpark of 20dB? I'm OK with small feedback cap, as my card will anyway be performing low cut at 90Hz.
Thanks bob. I've not managed to find that PCB here so went ahead and got my friend to rustle up one. No credit to me The PCB takes RN60 and Holco H4P plus Mills MRA-5. Other components need to be jury-rigged to fit. I wouldn't use this PCB anymore - basically you have to have the 3296 multiturn trimmers to get a decent adjustment. One of the changes I will be making now, I think the 3296 can be jammed in with some creativity.
Thanks bob. I've not managed to find that PCB here so went ahead and got my friend to rustle up one. No credit to me
The PCB takes RN60 and Holco H4P plus Mills MRA-5. Other components need to be jury-rigged to fit. I wouldn't use this PCB anymore - basically you have to have the 3296 multiturn trimmers to get a decent adjustment. One of the changes I will be making now, I think the 3296 can be jammed in with some creativity.
Member
Joined 2009
Paid Member
Umm, no chance of finding old IRF devices in my neck of the woods.
The ones I have used for this specific set of three boards were purchased from a very dodgy Chinese eBay shop, but they seem to work all right. Besides, they would be running at <500mA so I don't think they will be under pressure even if not original.
For my massive F5T build - pending for three years now - I have original IRFP240/9140 sourced from the local IR representative. Those will have to do.
As you can tell from the dialogue, my problem is actually at the front end as of now.
Need some help again...
I changed the values of the resistors to the suggested 150R/22R. Also went back to a single pair of input devices.
I have 7mA flowing through the front end biasing resistors (2.7k in my case) but 4mA of that is now flowing in the feedback network. Only 3mA is left over for the output stage biasing, which is not allowing me to increase bias beyond 3.5V.
As I see it I could either decrease the biasing resistor value or increase the collector resistor. Or increase the supply voltage (19V in the current situation), which is not an option presently.
I have 19.15 volts across the emitter feeding resistor, .615 across the 150R feedback resistor and .156 across the 22R emitter resistor. When maxing out the pots, I get 3.5V across G-S pins of output MOSFETs, nothing connected at input or output as of now. Behaviour is identical across two channels and both halves of the input stage.
Would appreciate any insight. I shall try to get the biasing current up to about 12mA, to adequately supply both the feedback network and the BJTs. However I'm a bit worried as I've never seen this kind of behaviour in any variant of the F5 I built so far. All the current is usually contained in the FE itself.
I changed the values of the resistors to the suggested 150R/22R. Also went back to a single pair of input devices.
I have 7mA flowing through the front end biasing resistors (2.7k in my case) but 4mA of that is now flowing in the feedback network. Only 3mA is left over for the output stage biasing, which is not allowing me to increase bias beyond 3.5V.
As I see it I could either decrease the biasing resistor value or increase the collector resistor. Or increase the supply voltage (19V in the current situation), which is not an option presently.
I have 19.15 volts across the emitter feeding resistor, .615 across the 150R feedback resistor and .156 across the 22R emitter resistor. When maxing out the pots, I get 3.5V across G-S pins of output MOSFETs, nothing connected at input or output as of now. Behaviour is identical across two channels and both halves of the input stage.
Would appreciate any insight. I shall try to get the biasing current up to about 12mA, to adequately supply both the feedback network and the BJTs. However I'm a bit worried as I've never seen this kind of behaviour in any variant of the F5 I built so far. All the current is usually contained in the FE itself.
Please attach exact schematic with components values and voltages that you got.