Well, I should mention that this is a proven design. It is only very unusual, and so many ways to make mistakes in the assembly process.
It is much easier to assemble something you designed yourself because you have sort of absorbed it. You sort of know where each part goes, but try to assemble something you don't know from just the instructions is much more error prone.
Jan
It is much easier to assemble something you designed yourself because you have sort of absorbed it. You sort of know where each part goes, but try to assemble something you don't know from just the instructions is much more error prone.
Jan
Same here, about 5MHz. Seems to be not self-oscillating but picked up from the environment. Persists after shutting down amp power. We do need anechoic chambers for our measurements
Hi BC,
Actually the 6MHz oscillation was 60MHz but has been fixed. I built up the prototype with MJE15032/33 output devices rather than the ones Winfield used, the 2SA1895 and 2SC4883. That wasn't such a good idea as it also killed the bandwidth.
There are two fixes to the PCB in addition to the inputs to the DC buck converter. One is about a clamp diode that has one connection in error, and a change in one connection of a comp cap. Not an error but it improves the frequency response.
The compensation pot let me nicely tune out any tendencies for instability.
I have also redrawn the basic input- and driver stages for better readability. See attached.
I now have retrofitted the recommended output devices and all is well. I need to update the 'known issues' doc at my website, but basically I am done.
I have boards left, not sure how many of the custom mounting plate I still have (which was a HUGE help to make a compact unit).
I will let you know Monday when I get back home.
Jan
Actually the 6MHz oscillation was 60MHz but has been fixed. I built up the prototype with MJE15032/33 output devices rather than the ones Winfield used, the 2SA1895 and 2SC4883. That wasn't such a good idea as it also killed the bandwidth.
There are two fixes to the PCB in addition to the inputs to the DC buck converter. One is about a clamp diode that has one connection in error, and a change in one connection of a comp cap. Not an error but it improves the frequency response.
The compensation pot let me nicely tune out any tendencies for instability.
I have also redrawn the basic input- and driver stages for better readability. See attached.
I now have retrofitted the recommended output devices and all is well. I need to update the 'known issues' doc at my website, but basically I am done.
I have boards left, not sure how many of the custom mounting plate I still have (which was a HUGE help to make a compact unit).
I will let you know Monday when I get back home.
Jan
Same here, about 5MHz. Seems to be not self-oscillating but picked up from the environment. Persists after shutting down amp power. We do need anechoic chambers for our measurements
I've had something similar with another project, turned out to be the scope ...
Jan
My bad, sorry, should indeed be 1360. The BOM is correct, thank Ohm.
Edit: posted the corrected redraw at Winfield Hill's 100W, 10MHz, 1000V/uS AMP-70 | Linear Audio NL
Jan
Edit: posted the corrected redraw at Winfield Hill's 100W, 10MHz, 1000V/uS AMP-70 | Linear Audio NL
Jan
Last edited:
Dear Jan,
great project! Do you have any PCBs left?
I am considering building a pair in bridge config for testing high-level amplitude response of a attenuator design of mine.
Maybe I should wait for your findings on BOM-details and tracking down of th oscillations...
I have two heatsink plates left, and several more PCBs. If you let me know (PM is OK) what you'd want and where to ship I can make a deal.
Jan
I have completed another of these Winfield Hill amps and did some functional tests. As you may remember, there is a trimmer on the board to adjust compensation to kill any oscillation.
After adjustment, another 'squegging' type oscillation was still present. This was conquered with base stoppers for the Vas cascode transistors.
All is described in detail in the List of Known Issues file which I have just updated.
Winfield Hill's 100W, 10MHz, 1000V/uS AMP-70 | Linear Audio NL
Jan
After adjustment, another 'squegging' type oscillation was still present. This was conquered with base stoppers for the Vas cascode transistors.
All is described in detail in the List of Known Issues file which I have just updated.
Winfield Hill's 100W, 10MHz, 1000V/uS AMP-70 | Linear Audio NL
Jan
No not yet, I had it on a 50 ohms load and quickly swept the signal generator and it looked level up to 5 or 6 MHz. I'm busy with stress projects atm so this will have to wait for later.
I did notice a DC offset of several 100mV though. I did increase the gain from 5 to 10 (power stage) by doubling the feedback R to 10k.
Jan
I did notice a DC offset of several 100mV though. I did increase the gain from 5 to 10 (power stage) by doubling the feedback R to 10k.
Jan
Depending on how you set the comp trimmer, about 3dB IIRC.
In my build I have removed the DC-DC converters and supply with analog floating small supplies (the main supply is a Hypex +/-46V SMPS) because I thought the 'squegging' was the DC-DC converters switching. Not so.
But it is eery just turning up the input freq beyond several MHz and the amp just keeps on going ... ;-)
A 100kHz square wave still looks like a square wave.
Jan
Sorry Jan for the late reply, work kept me buisy the last months.
Anyway. I have a question regarding the 1R Output. The Schematic states that it should not be used....
If I built this amp, I'll put it in a 2HE rack enclosure. Does it make sense to have only one 50R BNC output? I'd prefer having both, 1R on binding posts and a 50R output on coax.
Any ideas?
My remarks on the schematic about the choice of output resistors should be taken with a big grain of salt. Given the 5 to 10MHz bandwidth of the amplifier, it will benefit from a small series resistor, like 0.2 to 1 ohm, to isolate the effect of capacitive loads near its final rolloff. But other than that, it's really a matter of your application.
Yes, a 50-ohm resistor provides a back termination for driving a coax cable, and could dissipate as much as 20 watts into a 50-ohm load, so it's a serious issue, but it's a rather artificial issue as well.
The amplifier is fairly stable into all kinds of loads: I used it with the 1-ohm output to drive a step-up transformer to create several 100s of volts at 2.5MHz, into a capacitive load of coax and electrodes. The amplifier ended up being substantial overkill for that application.
Consider, if 1 ohm has 4 amps flowing through it, it could dissipate 16 watts, not an attractive idea. So for some applictions you'll want to use much smaller values.