This is turning into the evening of odd problems 
. I finally have some time to do some testing, and things are going strange all over the place.
I have built a headphone amp using an op amp driving a LT1206. I've done some distortion measurements, and I get 0.0027% THD. Nothing wrong with this, but I expected better.
After letting it run for a while I grabbed the LT1206 to see how hot it was getting (just warm), and noticed that the THD figure dropped to 0.0005%. Quite an improvement! I took my fingers away, and the THD went back up. The TO220 tab is isolated from the heat sink, so this only happens if I grab the LT1206 itself.
The same improvement occurs if I touch the tab with a multimeter probe, even with the meter off.
Any ideas as to what's happening here?
. I finally have some time to do some testing, and things are going strange all over the place. I have built a headphone amp using an op amp driving a LT1206. I've done some distortion measurements, and I get 0.0027% THD. Nothing wrong with this, but I expected better.
After letting it run for a while I grabbed the LT1206 to see how hot it was getting (just warm), and noticed that the THD figure dropped to 0.0005%
. Quite an improvement! I took my fingers away, and the THD went back up. The TO220 tab is isolated from the heat sink, so this only happens if I grab the LT1206 itself.The same improvement occurs if I touch the tab with a multimeter probe, even with the meter off.
Any ideas as to what's happening here?
There is 2200uF, 1uF, and 100nF to ground on each rail right next to the IC with a decent ground trace, as well as 100nF across the V+ and V- pins right under the IC.
No it isn't. The LT1206 tab is at V+ potential, and I didn't want to risk a short. Is it worth trying to ground the heat sink?
Last edited:
If the heat sink is not grounded, try grounding it and repeat the test.
There is capacitive coupling between the device and the heat sink.
Your amp may be oscillating because it’s seeing a capacitive load - try inserting a resistor in series with the load - put the resistor close to the opamp output.
Another possible cause is stray capacitance from the inverting nose to ground. At HF the gain increases and this may manifest as instability and/or noise.
Is your opamp unity gain stable?
Another possible cause is stray capacitance from the inverting nose to ground. At HF the gain increases and this may manifest as instability and/or noise.
Is your opamp unity gain stable?
Good way to check for oscillation (if you don't have a network analyzer) is just place an a.m. transistor radio near the offending circuit. Suggested by Analog Devices decades ago!
Hi BuckarooBanzai,
An oscilloscope, spectrum analyzer or network analyzer would be perfect for this. I would opt for an oscilloscope in this case. However, Jack's suggestion of an AM radio might possibly show you it is oscillating.
Bonsai also had some excellent advice, something you should implement if you haven't done so yet.
#1, ground that heat sink!!!! Use standard TO-220 insulators and thermal compound with a bushing to attach the chip to the heat sink. It only has to be snug, not tight. Most folks starting out over tighten heat sink hardware. That can crack the actual die inside the physical package, cut through sil-pads or squash the bushing to the point where it shorts.
So that's everything in one post. Every answer you got was good information.
-Chris
An oscilloscope, spectrum analyzer or network analyzer would be perfect for this. I would opt for an oscilloscope in this case. However, Jack's suggestion of an AM radio might possibly show you it is oscillating.
Bonsai also had some excellent advice, something you should implement if you haven't done so yet.
#1, ground that heat sink!!!! Use standard TO-220 insulators and thermal compound with a bushing to attach the chip to the heat sink. It only has to be snug, not tight. Most folks starting out over tighten heat sink hardware. That can crack the actual die inside the physical package, cut through sil-pads or squash the bushing to the point where it shorts.
So that's everything in one post. Every answer you got was good information.
-Chris
I'd worry that you are linking everything together at HF - normally the 100nF is just between V+ and V- and is the only decoupling that needs to be right on the chip as its the only high speed decoupling.
Decoupling caps to ground have to be sited carefully as they can inject distorted half waveforms into the ground trace, though this isn't probably the issue you see, its more a power amplifier concern.
Perhaps there is some VHF coupling going on here with all the 100nF's joining the rails and ground?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.