No, such large RF-ringing 'blibs' are not normal.
Did you mod the design? Different layout? Adjustment of gatedrive & dead time?
Or is the scope probe connected with a long GND wire forming a loop partially over the switching stage?
Nothing modified at all, same PCB, (Your PCB) scope GND is connected directly to the GND pin on the PCB.
For double checking I grabbed my old sample (+/-40V version) from my storage. Attached what I am getting.
Looks clean even when using the long GND clip in an ignorant way.
In your screen shot there is more which does not fit.
The frequency is high.
It is even above the max frequency of the +/-40V version, but at the same time the magnitude of the carrier looks more like a +/-80V version.
Could it be that some misplaced component in the area of the feedback or modulator is pushing your +/-80V version to higher frequencies?
If so then this could lead to such blips, because under this condition the ZVS during idle would be messed.
For your reference I am attaching my measurement:
Scope Tek TDS784C set to a sampling rate of 4Gs/s and full 1GHz bandwidth. Used probe was a 450MHz 1:10 probe (Elditest GE4512).
Scales of screen shot:
Time base: 400ns / Div.
White trace: Output voltage, 200mV/Div.
Looks clean even when using the long GND clip in an ignorant way.
In your screen shot there is more which does not fit.
The frequency is high.
It is even above the max frequency of the +/-40V version, but at the same time the magnitude of the carrier looks more like a +/-80V version.
Could it be that some misplaced component in the area of the feedback or modulator is pushing your +/-80V version to higher frequencies?
If so then this could lead to such blips, because under this condition the ZVS during idle would be messed.
For your reference I am attaching my measurement:
Scope Tek TDS784C set to a sampling rate of 4Gs/s and full 1GHz bandwidth. Used probe was a 450MHz 1:10 probe (Elditest GE4512).
Scales of screen shot:
Time base: 400ns / Div.
White trace: Output voltage, 200mV/Div.
Attachments
Last edited:
Hi Markus,
Due long inductor wires, that's why I had the problem, the inductor was made and wired quickly for testing purposes.
See below, Image with full 100MHz bandwidth DS0009, and Image with 20MHz bandwidth limit ON DS0010.
This is the +-40V Version, I am going to build the +-80V version next week.
Thanks.
Does it look better now?
Due long inductor wires, that's why I had the problem, the inductor was made and wired quickly for testing purposes.
See below, Image with full 100MHz bandwidth DS0009, and Image with 20MHz bandwidth limit ON DS0010.
This is the +-40V Version, I am going to build the +-80V version next week.
Thanks.
Does it look better now?
Attachments
Last edited:
From you screen shot I am guessing that you are working on a +/-80V version and most likely it is suffering from one or multiple wrong component values, which leads to a very high oscillation frequency of 456kHz, while the +/-80V is designed for roughly 260...310kHz.
The combination of 456kHz and carrier size in your screen shot indicates a wrong value in the output filter. Additional there might be wrong values in the modulator or feedback network.
Here a screen shot how it should look, when everything is right.
... took this picture for double checking on my +/-80V sample today.
Scope: Tektronix TDS784C set to a sampling rate of 4GS/s and full 1GHz bandwidth.
Probe: 450MHz 1:10 probe (Elditest GE4512).
Time base: 400ns / Div.
Signal trace: Output voltage, 500mV/Div.
Attachments
@ Choco
My supply is +-40V. output filter is 15uH. The other PCB I have will be assembled soon for the +-80V version.
I am going to check the values again, but that's very strange indeed.
My supply is +-40V. output filter is 15uH. The other PCB I have will be assembled soon for the +-80V version.
I am going to check the values again, but that's very strange indeed.
Time overlap during posting...
So simply use my last post for reference for the +/-80V type.
Just scratching my head about the size of the carrier residual.
When adjusted to 400kHz, the typical Pk-Pk size of the carrier residual at the output is 1.2Vpp..1.3Vpp. Your screen shot shows more than double of that, which would indicate still a wrong component value in the output filter.
So simply use my last post for reference for the +/-80V type.
Yes, this is looking much better 🙂
Just scratching my head about the size of the carrier residual.
When adjusted to 400kHz, the typical Pk-Pk size of the carrier residual at the output is 1.2Vpp..1.3Vpp. Your screen shot shows more than double of that, which would indicate still a wrong component value in the output filter.
15uH is fine.
Are C24+C25+C26 together summing up to approx 0.77uF?
400kHz switching frequency in combination with supply rail voltage +/-40V and output filter of 15uH / 0.77uF simply do not fit together with the size the signal in your screen shot.
Are C24+C25+C26 together summing up to approx 0.77uF?
400kHz switching frequency in combination with supply rail voltage +/-40V and output filter of 15uH / 0.77uF simply do not fit together with the size the signal in your screen shot.
Last edited:
Hello Chocoholic,
i want to build the 2x65V Version and drive it from my 2x58V supply, are there any components i need to change in the design?
Is there any chance to get in contact with you (since i work in Munich) about some wire for the inductor or any other parts you want to get rid of🙂 you could pm me your email. cheers!
i want to build the 2x65V Version and drive it from my 2x58V supply, are there any components i need to change in the design?
Is there any chance to get in contact with you (since i work in Munich) about some wire for the inductor or any other parts you want to get rid of🙂 you could pm me your email. cheers!
Shape is perfect. 
Also size is coming closer, but it still deviates by factor 1.6 vs. the expected size.
When tuned to 423kHz and rails +/-40V and C24+C25+C26=0.77uF and L=15uH then the expected Pk-Pk size is 1.15Vpk-pk.
Your screen shot would fit to C24+C25+C26=0.44uF, i.e. when C26 would be missing or not soldered.
Also size is coming closer, but it still deviates by factor 1.6 vs. the expected size.
When tuned to 423kHz and rails +/-40V and C24+C25+C26=0.77uF and L=15uH then the expected Pk-Pk size is 1.15Vpk-pk.
Your screen shot would fit to C24+C25+C26=0.44uF, i.e. when C26 would be missing or not soldered.
..enjoy your amp.. 🙂
Attached two screen shots with similar signal from my sample when operated from +/-39V without load.
Measurement with x10 probe ==> Vertical scale: 10V/Div.
Attachments
Back from a long break from diy and coming back to this ... with more questions I'm affraid.
What kind of offset do we expect from the amp itself ? Is the offset correction in the gain stage only intended to correct errors in the two opamps inamp or for the whole amplifier ?
Thanks in advance 🙂
What kind of offset do we expect from the amp itself ? Is the offset correction in the gain stage only intended to correct errors in the two opamps inamp or for the whole amplifier ?
Thanks in advance 🙂
At the moment I am working on a variant with the IR4301M to make an ultra small companion.
Just need a little more time to finish it (also because I am kind of traveling/moving right now)
Just need a little more time to finish it (also because I am kind of traveling/moving right now)
The offset correction on the gain board allows to compensate all the offsets throughout the entire LiteAmp to the speaker. So you can nicely adjust it to zero.
If you run the main amp without gain board and use an input coupling cap then you can expect DC errors below 20mV.
If you run the +/-40V main board without gain board and drive it with DC coupling you have to expect the following DC voltage at the output:
VDC = -4.9 * (DC error of chosen preamp + (+/- 20mV))
