After following this forum for a while I though I would attempt to make my own UCD. Attached is my non-discrete design I have come up with. Please comment on my design where you see fit.
I plan to run the circuit off +/- 20 V supplies to with Vcc=-Vee=12. Most of the parts I have chosen for avaibility and price as it is expensive to aquire not locally avaible components in New Zealand.
Input buffer opamps: NE5532D
Comparators: LM319M
H-Bridge driver: LM5101
MOSFETS: FQD20N06
Also importantly, I am not quite sure how to calculate the values of the feedback network components correctly.
I understanding of the theory in that the oscillation frequency is that at which the phase shift from the modulator, output filter and feedback is 180 degrees.
Would it be good to increase the values of C18 & C20 to slow the comparator response time in order to get a reasonable delay and oscialtion frequency. I think the comparators and gate drivers response time of 80ns and 25ns respectivly is almost too fast.
Cheers for any input, Merry Christmas,
George
I plan to run the circuit off +/- 20 V supplies to with Vcc=-Vee=12. Most of the parts I have chosen for avaibility and price as it is expensive to aquire not locally avaible components in New Zealand.
Input buffer opamps: NE5532D
Comparators: LM319M
H-Bridge driver: LM5101
MOSFETS: FQD20N06
Also importantly, I am not quite sure how to calculate the values of the feedback network components correctly.
I understanding of the theory in that the oscillation frequency is that at which the phase shift from the modulator, output filter and feedback is 180 degrees.
Would it be good to increase the values of C18 & C20 to slow the comparator response time in order to get a reasonable delay and oscialtion frequency. I think the comparators and gate drivers response time of 80ns and 25ns respectivly is almost too fast.
Cheers for any input, Merry Christmas,
George
Hi. It looks like an interesting project. An IC version of UCD can be made very cheep, high performance and best of all simple with respect to the number of components on a PCB.
The circuit as it is now will probably work, I do have some suggestions to it though.
I would recomend you to place a large resistor in parralel with C13 and C14, to prevent a large pulse at the output, when you attach an input device.
It doesn't look like LM5101 prevents the two MOSFET's from beeing turned on at the same time. You have to make sure that the comperators detect a crossover at the same time. If not, you could get problems with distortion or worse simultaneous conduction -> failure of the MOSFET's. If you have access to LM5104 with a single input, you would be on the safe side. Just don't use LM5104 in a H-bridge configuration, I've tryed it. It's so sensitive to noise, that it won't work (it's most likely the pulse driven high side driver, that haven't got sufficient noise rejection).
The whole instrumental amplifier configuration is nearly floating. R1 and R2 provides some DC operation point at startup, and later the feedback from output will help. If, somehow the dynamic range of the comperators is exceeded, the MOSFET's could turn on simultaneously.
I don,'t know how this floting configuration will affect the sound quality, but I guess you got it from Bruno Putzeys papers on UCD, so I wouldn't argue against it.
Since LM319 has an open drain configuration, it will turn low much faster than high, which will give you deedtime (some is needed) and provide some safety margin against simultaneously conduction.
NE5532D has some nice specifications.
For a +/-20V operation, you could choose a MOSFET with a lower on resistance, without the penalty of high losses due to the switching of the consequently larger output capacitance of the MOSFET's (P=1/2*C*U^2*f). This will bring down the total MOSFET losses. For now FQD20N06 will probably be just fine.
The peak diode recovery dv/dt and the overall ruggedness Farichild's MOSFET's is not that good. You might experince failure due to this. Keep that in mind, if you experince unaccountable MOSFET failures. IRF produce more rugged devices with good Ron, Cgd, Coss specifications.
About the propagation delay. ~100ns sounds fine. Keep in mind that you already have a phaseshift of -180 at ~40kHz from the output filter, so you just need the smallest extra phase shift from the delay, to start the osccilation. The purpose of C30 is to provide some positive phaseshift up to somewhere arround 300-400kHz where you really want the osccilation. At this frequency the delay of 100ns has already caused an additional phase shift of -10 degress (-20 with 200ns delay). so you will get you oscillation. You can lower the oscillation frequency by incressing R20 nd R18. If the oscillation takes place a arround 50kHz increese C30 and C29.
The size of C18 and C20 works as a capacitive voltage divider together with C29 and C30. I don't see how C18 and C20 affects the propagation delay. The size of C18 and C20 together with R18 and R20however gives you some extra negative phase shift, which wil help you attain oscillation.
The circuit as it is now will probably work, I do have some suggestions to it though.
I would recomend you to place a large resistor in parralel with C13 and C14, to prevent a large pulse at the output, when you attach an input device.
It doesn't look like LM5101 prevents the two MOSFET's from beeing turned on at the same time. You have to make sure that the comperators detect a crossover at the same time. If not, you could get problems with distortion or worse simultaneous conduction -> failure of the MOSFET's. If you have access to LM5104 with a single input, you would be on the safe side. Just don't use LM5104 in a H-bridge configuration, I've tryed it. It's so sensitive to noise, that it won't work (it's most likely the pulse driven high side driver, that haven't got sufficient noise rejection).
The whole instrumental amplifier configuration is nearly floating. R1 and R2 provides some DC operation point at startup, and later the feedback from output will help. If, somehow the dynamic range of the comperators is exceeded, the MOSFET's could turn on simultaneously.
I don,'t know how this floting configuration will affect the sound quality, but I guess you got it from Bruno Putzeys papers on UCD, so I wouldn't argue against it.
Since LM319 has an open drain configuration, it will turn low much faster than high, which will give you deedtime (some is needed) and provide some safety margin against simultaneously conduction.
NE5532D has some nice specifications.
For a +/-20V operation, you could choose a MOSFET with a lower on resistance, without the penalty of high losses due to the switching of the consequently larger output capacitance of the MOSFET's (P=1/2*C*U^2*f). This will bring down the total MOSFET losses. For now FQD20N06 will probably be just fine.
The peak diode recovery dv/dt and the overall ruggedness Farichild's MOSFET's is not that good. You might experince failure due to this. Keep that in mind, if you experince unaccountable MOSFET failures. IRF produce more rugged devices with good Ron, Cgd, Coss specifications.
About the propagation delay. ~100ns sounds fine. Keep in mind that you already have a phaseshift of -180 at ~40kHz from the output filter, so you just need the smallest extra phase shift from the delay, to start the osccilation. The purpose of C30 is to provide some positive phaseshift up to somewhere arround 300-400kHz where you really want the osccilation. At this frequency the delay of 100ns has already caused an additional phase shift of -10 degress (-20 with 200ns delay). so you will get you oscillation. You can lower the oscillation frequency by incressing R20 nd R18. If the oscillation takes place a arround 50kHz increese C30 and C29.
The size of C18 and C20 works as a capacitive voltage divider together with C29 and C30. I don't see how C18 and C20 affects the propagation delay. The size of C18 and C20 together with R18 and R20however gives you some extra negative phase shift, which wil help you attain oscillation.
Cheers for the feedback sovadk and drzhuang.
I have made a few changes and reattached my schematic. Mainly to do with the way I have been using the comparators. Allowing C13 & C14 to discharge is also important yes.
Hopefully with the comparators being manufactured on the same die and the complementary situation that they are in I will not have trouble with cross conduction or a large deadtime.
Thanks for pointing this out. That was a big mistake. I did select the LM319 as it does provide a means to reference its output and I have now utilized this to provide the level shifting. I am pretty sure I should be operating within the limits of the LM319. If this doesn't work I will have to add some level shifting mechanism. Operation at higher voltages would also probably require discrete transistor level shifting on the output of the comparator.
This is quite a big learning experiance for me as I am still a student and I wont be surprised if things dont work first time. Tomorrow I will order the components and design a PCB.
Cheers, George.
I have made a few changes and reattached my schematic. Mainly to do with the way I have been using the comparators. Allowing C13 & C14 to discharge is also important yes.
Hopefully with the comparators being manufactured on the same die and the complementary situation that they are in I will not have trouble with cross conduction or a large deadtime.
Thanks for pointing this out. That was a big mistake. I did select the LM319 as it does provide a means to reference its output and I have now utilized this to provide the level shifting. I am pretty sure I should be operating within the limits of the LM319. If this doesn't work I will have to add some level shifting mechanism. Operation at higher voltages would also probably require discrete transistor level shifting on the output of the comparator.
This is quite a big learning experiance for me as I am still a student and I wont be surprised if things dont work first time. Tomorrow I will order the components and design a PCB.
Cheers, George.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.