How low can you set the gain, I would like to try this as a headphoneamp so I don't need more than a gain of 3 or so.
I have honestly no idea.
You can always have a try, and if you are confronted with instabilities, you can add local compensation capacitors between the (-) input and the output of each opamp.
Something in the pF's or tens of pF range should do.
Cool! How's it look at more typical output powers (say, 100uW to 10mW) and more difficult frequencies (say, 10-20kHz)?
The OPA2134 family could work; pretty decent audio op amp with 8MHz GBP typ. Among the National parts the LME49723 (17MHz GBP typ) is the slowest that'd take +-12V.
Thanks for sharing this unusual amplifier. I'm collecting parts for a breadboard version. How much current do I need from the power supply to get it up and running for testing? I've got a +/- 15V power supply from an old project that I'd like to use.
With the 1.5Ω resistor in the middle regulator, the current drain will be ~850mA.
Round to 1 amp to be sure.
The resistor could be increased to reduce the output power, or to drive higher impedance loads.
Note that this is class A, and the power has to be supplied even when the output power is 0 (and all of that power has to be dissipated by the regulators, which means good heatsinks are mandatory).
After putting together Elvee's amplifier, using the parts I had at hand and listening to it (I had a pair of full range car speakers at hand, but 4 ohm impedance), I decided this deserves a nice little layout and further testing on some better speakers.
So far I finished the layout (a double layer design) and I'm in the process of making the actual PCB. Once I'll have the prototype up and running, I'll post a zip archive with the complete eagle package, some photos of the prototype and the BOM. I used Panasonic FC elcos for this test, except for the 2200uF output caps that are from RIFA (Kemet actualy... I learned that they bought RIFA).
I did not pay attention to the resistors at this point... I think they're 1% metal film type.
He're a quick preview of the schematic and layout... There is still some work left to do, like component values and so on...
The SE_class_A_brd.pdf has both top and bottom layers and the one with _bot at the end is just the bottom layer, with the components still visible...
Tell me what you think!
So far I finished the layout (a double layer design) and I'm in the process of making the actual PCB. Once I'll have the prototype up and running, I'll post a zip archive with the complete eagle package, some photos of the prototype and the BOM. I used Panasonic FC elcos for this test, except for the 2200uF output caps that are from RIFA (Kemet actualy... I learned that they bought RIFA).
I did not pay attention to the resistors at this point... I think they're 1% metal film type.
He're a quick preview of the schematic and layout... There is still some work left to do, like component values and so on...
The SE_class_A_brd.pdf has both top and bottom layers and the one with _bot at the end is just the bottom layer, with the components still visible...
Tell me what you think!
It looks good 🙂
Maybe you could include the input blocking caps?
Since the output is AC coupled anyway, there is no reason to keep the DC at the input, where it could upset the bias voltages.
Well, I just etched the PCB today, so I'm going to do some tests with and without DC blocking caps. This would require a slight change of the input stage.
But, for sure this is not the final version of the PCB, so the next step will be this: including DC blocking caps at the input.
What I'm a little worried about with this layout is the ground concept... Can't wait until tomorrow to place the components and fire it up!
But, for sure this is not the final version of the PCB, so the next step will be this: including DC blocking caps at the input.
What I'm a little worried about with this layout is the ground concept... Can't wait until tomorrow to place the components and fire it up!
Yeah... I got a set-back today... I seem to have misplaced my samples of LME49880 and I have no other audio op-amps at hand (with SO8 package that is...) 🙁.
So... I'll have to wait a little bit until I get new parts (or find the misplaced ones...).
So... I'll have to wait a little bit until I get new parts (or find the misplaced ones...).
At laassst! My greatest work! Is compleeeeted!
Well, yeah, today I powered up the amplifier for the first time, and IT WORKS!
I works quite nicely, almost dead silent with the input cable connected to the PC's headphone output. Just a slight hiss is heard when you are very close to the speaker.
I only had a pair of 4 ohm car speakers to test the amp, but it sounds very nicely. The speakers were on the table, not mounted in an enclosure, so I could not judge the bass quite well. But the highs are very nice. I played a couple of tracks (both mp3 and flac) and I'm happy with it (so are two of my colleagues who share the same passion for DIY audio).
Hopefully tomorrow I'll have time to post some pictures of the assembled amp and soon some oscilloscope screenshots at different frequencies. Maybe a test with RMAA...
Keep in touch!
Well, yeah, today I powered up the amplifier for the first time, and IT WORKS!
I works quite nicely, almost dead silent with the input cable connected to the PC's headphone output. Just a slight hiss is heard when you are very close to the speaker.
I only had a pair of 4 ohm car speakers to test the amp, but it sounds very nicely. The speakers were on the table, not mounted in an enclosure, so I could not judge the bass quite well. But the highs are very nice. I played a couple of tracks (both mp3 and flac) and I'm happy with it (so are two of my colleagues who share the same passion for DIY audio).
Hopefully tomorrow I'll have time to post some pictures of the assembled amp and soon some oscilloscope screenshots at different frequencies. Maybe a test with RMAA...
Keep in touch!
Congrats!!
4 ohm is a bit tough on this amplifier: since the voltage regs are limited to 1.5A, this severely restricts the available power on low impedance loads
So how would performance change by moving to a LT1964 in place of the lm317?
http://cds.linear.com/docs/Datasheet/1963fc.pdf
http://cds.linear.com/docs/Datasheet/1963fc.pdf
The LT1963 is a LDO, and this is a fundamental difference.
I think it would at the very least require a complete makeover of the compensation scheme.
I do not even know if stability is achievable with a LDO.
In addition, a complementary (negative) regulator would have to be found, and I am not sure it would be easy.
I think it would at the very least require a complete makeover of the compensation scheme.
I do not even know if stability is achievable with a LDO.
In addition, a complementary (negative) regulator would have to be found, and I am not sure it would be easy.
Would an LM338T be a suitable replacement for the LM317T if a bit more power was required? The specs look very similar, and anything that mentions frequency (ripple rejection etc.) seem close too.
Brian.
Brian.
Yes, but there is the problem of the negative counterpart: I don't think there is one.
Good point Elvee. So, it would be possible to build two of the top half of the circuit, but you would lose the efficiency of correlation. I suppose you could then get a DC coupled out stage though... <Goes looking for big heat sinks>
Thanks,
Brian.
Thanks,
Brian.
I was wondering if anyone's tested the overload/recovery behaviour?
If the top LM317 goes into thermal shutdown, it seems like the string of regulators would all loose power and shut down, then the opamps would go nuts trying to bang through the broken feedback loop. If we stick a relatively large value resistor (say 100ohms) between Adj and Out, would this patch up the issue during the thermals shutdown until the regulators turn back on?
Another look at this, it seems if the driver opamps maintained the voltage difference between the current source in the middle while top of bottom shuts down, it would try to draw a pretty crazy amount of current from the opamps, and I am not so sure how the opamp would react in this sort of situation. (A large voltage drop across the 100R feedforward resistor?)
If the top LM317 goes into thermal shutdown, it seems like the string of regulators would all loose power and shut down, then the opamps would go nuts trying to bang through the broken feedback loop. If we stick a relatively large value resistor (say 100ohms) between Adj and Out, would this patch up the issue during the thermals shutdown until the regulators turn back on?
Another look at this, it seems if the driver opamps maintained the voltage difference between the current source in the middle while top of bottom shuts down, it would try to draw a pretty crazy amount of current from the opamps, and I am not so sure how the opamp would react in this sort of situation. (A large voltage drop across the 100R feedforward resistor?)
Last edited: