Thought I would share a project I've been working on that is an adaptation of the THAT1646-based class-A headphone amp I developed some time ago.
Adding larger transistors and heatsinks permit the class-A headphone amp board to deliver about 10W of class-A power into 4 or 8 Ohms. The quiescent current is a whopping 1.25A/channel so I explored the use of low-cost 18V 60W (3.33A) Meanwell laptop switchers.
This is a picture of the stereo amp test-bed. The design is fully DC-coupled and does not require a servo. The output runs open loop and is stable with any load capacitance with the maximum value limited by output current.
Output Vos is typically 6 mV or less. Heatsink pictures show a smaller BD139 used to sense temperature and provide thermal feedback to the DRV134's differential input.
The DRV134 uses "common mode drive" via its sense pins to provide AC drive to the output devices. The DRV134 "differential drive" input mirrors the Vbe of the sense transistor to provide both a positive and negative Vbe for the output devices. The design is remarkably thermally stable.
The DRV134 and external output stage are unity gain so there's not a lot of noise: It seems silly to have an amp with 25-30 dB of voltage gain preceded by an attenuator so I can listen at the 100 mW level. But, anyone using this will likely need a gain stage ahead of it. I intend to use an active control similar to Self's designs.
Here are some measurements I ran before I had the switching supplies. The idle current is only about 910 mA rather than 1.2A so the 8W THD figures will be somewhat lower.
8 mW into 8 Ohms Iq 910 mA no analyzer attenuation
80 mW into 8 Ohms Iq 910 mA no analyzer attenuation
660 mW into 8 Ohms Iq 910 mA
8W into 8 Ohms Iq 910 mA
You can follow the design here: Pro Audio Design Forum • View topic - A Simple 10W Direct-Coupled Class-A Power Amplifier
The headphone amp project is here: http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=7&t=509
This is an amp I'm going to use on my bench: I've been blaming my speakers for everything sounded gritty at low level and it turns out it was the power amps (more than one) I was using.
I've been several decades late arriving to the conclusion that class-A sounds better. Maybe its because I don't listen at the levels I used to.
Enjoy;
Wayne
Adding larger transistors and heatsinks permit the class-A headphone amp board to deliver about 10W of class-A power into 4 or 8 Ohms. The quiescent current is a whopping 1.25A/channel so I explored the use of low-cost 18V 60W (3.33A) Meanwell laptop switchers.
This is a picture of the stereo amp test-bed. The design is fully DC-coupled and does not require a servo. The output runs open loop and is stable with any load capacitance with the maximum value limited by output current.
Output Vos is typically 6 mV or less. Heatsink pictures show a smaller BD139 used to sense temperature and provide thermal feedback to the DRV134's differential input.
The DRV134 uses "common mode drive" via its sense pins to provide AC drive to the output devices. The DRV134 "differential drive" input mirrors the Vbe of the sense transistor to provide both a positive and negative Vbe for the output devices. The design is remarkably thermally stable.
The DRV134 and external output stage are unity gain so there's not a lot of noise: It seems silly to have an amp with 25-30 dB of voltage gain preceded by an attenuator so I can listen at the 100 mW level. But, anyone using this will likely need a gain stage ahead of it. I intend to use an active control similar to Self's designs.
An externally hosted image should be here but it was not working when we last tested it.
Here are some measurements I ran before I had the switching supplies. The idle current is only about 910 mA rather than 1.2A so the 8W THD figures will be somewhat lower.
8 mW into 8 Ohms Iq 910 mA no analyzer attenuation
80 mW into 8 Ohms Iq 910 mA no analyzer attenuation
660 mW into 8 Ohms Iq 910 mA
8W into 8 Ohms Iq 910 mA
You can follow the design here: Pro Audio Design Forum • View topic - A Simple 10W Direct-Coupled Class-A Power Amplifier
The headphone amp project is here: http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=7&t=509
This is an amp I'm going to use on my bench: I've been blaming my speakers for everything sounded gritty at low level and it turns out it was the power amps (more than one) I was using.
I've been several decades late arriving to the conclusion that class-A sounds better. Maybe its because I don't listen at the levels I used to.
Enjoy;
Wayne
Last edited:
Interesting design - I've not seen the DRV134 used in common-mode before.
I do have a question though - with a typical HFE of the output devices around 100 or so and an idle current of 1A+ it does look as though you're operating the DRV134 in classAB even though the output stage is in classA. Have you considered biassing the DRV with current sources on the outputs to 'purify' the amp's classA credentials?
I do have a question though - with a typical HFE of the output devices around 100 or so and an idle current of 1A+ it does look as though you're operating the DRV134 in classAB even though the output stage is in classA. Have you considered biassing the DRV with current sources on the outputs to 'purify' the amp's classA credentials?
Thanks for the reply.
Roger Foote used this design in his P3SME mastering compressor as a transformer driver. It's turned out to be a very useful circuit. I started playing around with common mode drive of the THAT1646 some time ago: http://www.proaudiodesignforum.com/forum/php/viewtopic.php?f=6&t=140
To be honest I haven't looked at current source loading the DRV134 outputs.
Last edited:
Looking at how the distortion rises - in particular the higher order harmonics - at higher output levels I'm curious whether those would improve with CCS loading.
Member
Joined 2009
Paid Member
I would not expect that to make much of a difference. I suspect the distortion is from the output power devices running open loop. If it sounds good, I wouldn't change a thing.
You're never curious even when something sounds good that maybe something could be changed to get it to sound even better? My curiosity's been rewarded like that on more than one occasion....😛
I'm pretty sure that the output stage was under-biased with an Iq of 910 mA at the 8W level. 910 mA would probably be sufficient for about 6W out into 8R. I'm currently running it at 1.2A/channel Iq. (0.75x1.6Apk)
I think what were seeing is compressive output stage distortion at Iq 910 mA and 8W output. Some of it is also IG-18 generator at the top of its output range.
The DRV134 and THAT1646 on their own don't produce that kind of signature. I've run a a lot of high-level highly-capacitive-loaded tests on them as line drivers and it is uncharacteristic.
I think what were seeing is compressive output stage distortion at Iq 910 mA and 8W output. Some of it is also IG-18 generator at the top of its output range.
The DRV134 and THAT1646 on their own don't produce that kind of signature. I've run a a lot of high-level highly-capacitive-loaded tests on them as line drivers and it is uncharacteristic.
I reckon (just a gut inclination no more) that when the OPS starts going out of classA the input currents start to make it look like a very non-linear load. The DRV's output Rs (50R) are going to end up with fairly nasty-looking haversine voltages imposed across them and that could be the cause of all the higher-order stuff.
I suspect that you're right about the role of the DRV134's 50 Ohm output resistors once the OPS drops out of class-A.
Having said that 8W/910mA Iq is right at the OPS class-A drop-out. Below that THD drops very rapidly. The idle current is directly related to that point. I suppose I should have posted FFTs of it near 10W with an Iq of 1.2A rather than 910 mA but that's what I had.
When I first tried it the Iq was 85 mA (small devices tiny heatsinks for HPA use) and the THD was in the 3-10% range at high levels driving 8 Ohms. And it still sounded good because the distortion occurs in compressive form at signal peaks. Low-level detail isn't affected.
I'm not gonna complain about the 0.04% THD at 8W when it was under-biased by 25-30%.
Having said that 8W/910mA Iq is right at the OPS class-A drop-out. Below that THD drops very rapidly. The idle current is directly related to that point. I suppose I should have posted FFTs of it near 10W with an Iq of 1.2A rather than 910 mA but that's what I had.
When I first tried it the Iq was 85 mA (small devices tiny heatsinks for HPA use) and the THD was in the 3-10% range at high levels driving 8 Ohms. And it still sounded good because the distortion occurs in compressive form at signal peaks. Low-level detail isn't affected.
I'm not gonna complain about the 0.04% THD at 8W when it was under-biased by 25-30%.
I did a quick FFT to measure the THD with an idle current of 1.2A.
Unlike the previous measurement the generator was an analog IG-18.
The green channel is generator monitor.
The red channel is 8W into 8 Ohms with an Iq of 1.2A.
8W is approx 1.5 dB below the clip point with 18V supplies.
The 1 kHz THD is about 0.019%.
(Note: There is a 26 dB pad at the analyzer input. The measurement is 24 bit, 96 kHz so the noise floor is elevated by the A/D and lack of meaningful shielding.)
A Simple 10W Direct-Coupled Class-A Amp Using The DRV134.
FFT of output at 8W into 8 Ohms with an Iq of 1200mA.
This test was performed with the amp powered by two Meanwell 60W switchers.
Unlike the previous measurement the generator was an analog IG-18.
The green channel is generator monitor.
The red channel is 8W into 8 Ohms with an Iq of 1.2A.
8W is approx 1.5 dB below the clip point with 18V supplies.
The 1 kHz THD is about 0.019%.
(Note: There is a 26 dB pad at the analyzer input. The measurement is 24 bit, 96 kHz so the noise floor is elevated by the A/D and lack of meaningful shielding.)
A Simple 10W Direct-Coupled Class-A Amp Using The DRV134.
FFT of output at 8W into 8 Ohms with an Iq of 1200mA.
This test was performed with the amp powered by two Meanwell 60W switchers.
Last edited:
- Status
- Not open for further replies.