My first attempt at full range audio class D was on breadboard with IR2010 and LM311. I used +/-24V rails and IRF520. It worked without much trouble, so it's achievable, but you have to keep the power layout very tight.
Consider placing the MOSFET right in front of the driver and the supply capacitors very close on one side. Output inductor may be at a longer distance, but output capacitor shouldn't.
Use star grounding to the center of the supply capacitors and cut wire bridges of minimum length for power nodes. Use a breadboard in good condition, false contacts can kill the MOSFET and the driver instantaneously. The modulator may be far from the power stuff.
Remember ferrite beads on supply rails before they exit the breadboard, etc... It's achievable if you place everything properly. It will be much easier if you have experience laying out SMPS PCBs.
Consider placing the MOSFET right in front of the driver and the supply capacitors very close on one side. Output inductor may be at a longer distance, but output capacitor shouldn't.
Use star grounding to the center of the supply capacitors and cut wire bridges of minimum length for power nodes. Use a breadboard in good condition, false contacts can kill the MOSFET and the driver instantaneously. The modulator may be far from the power stuff.
Remember ferrite beads on supply rails before they exit the breadboard, etc... It's achievable if you place everything properly. It will be much easier if you have experience laying out SMPS PCBs.
The below URL is Mr.Kamijo's Web where he published
an RS2092 based D Class Amplifier whose raw PCB is
a threw-in with a Japanese Magazine named Transistor Gijyutsu.
http://www.ne.jp/asahi/evo/amp/DQamp/D9amp.htm
Many people used this PCB succsseffly. You can see from the
photo, MOSFET's and inductor for LPF are allocated near by
IRS2092 as well as two 2.200uF capacitors which are by-pass
for +/- rails in addition to the bulk capacitors near the transformer.
an RS2092 based D Class Amplifier whose raw PCB is
a threw-in with a Japanese Magazine named Transistor Gijyutsu.
http://www.ne.jp/asahi/evo/amp/DQamp/D9amp.htm
Many people used this PCB succsseffly. You can see from the
photo, MOSFET's and inductor for LPF are allocated near by
IRS2092 as well as two 2.200uF capacitors which are by-pass
for +/- rails in addition to the bulk capacitors near the transformer.
I am building an amp on breadboard (first) based on the IRS2092 and using IRFB4019PbF MOSFET's. The first circuit i built was using regulators to supply VAA and VSS. decided to omit the regulators and stick strictly with the typical application circuit shown the in the Application Note AN-1138. I am getting a PWM signal from the LO of the IRS2092, but nothing on HO.
Hi, Rob. I am sure that you already know this, if so, sorry: have in mind that in order to measure the HO output you need an oscilloscope with floating ground, with its ground connected to the VS node of your circuit, and the tip of the probe at HO. Don't expect a very clean waveform this way, but at least it will tell you roughly if you have something there.
Having said that, if you are getting output on LO (a square wave of around 12-15Vpp, referred to -Vbus), but nothing on HO, then you probably have no valid bootstrap voltage.
Check the VB voltage. As a first attempt, you can use a multimeter between VB and VS pins. It should be around 20V first, when the amplifier is in shutdown mode, and then go down to around Vcc volts (12 to 15V) and remain estable at this value. If this is not true, use the oscilloscope in the same way and check what's the waveform at VB referred to VS.
If you really have LO but no HO waveform, how does your VS waveform look like? I guess it seems ok, a square wave from -Vbus to Vbus, but then how is your output when you inject a sine wave? It may be almost ok at very low output but then starting to get severely distorted as you increase amplitude. Is this what you observe?
Tell us how these tests go in order to continue helping you.
Having said that, if you are getting output on LO (a square wave of around 12-15Vpp, referred to -Vbus), but nothing on HO, then you probably have no valid bootstrap voltage.
Check the VB voltage. As a first attempt, you can use a multimeter between VB and VS pins. It should be around 20V first, when the amplifier is in shutdown mode, and then go down to around Vcc volts (12 to 15V) and remain estable at this value. If this is not true, use the oscilloscope in the same way and check what's the waveform at VB referred to VS.
If you really have LO but no HO waveform, how does your VS waveform look like? I guess it seems ok, a square wave from -Vbus to Vbus, but then how is your output when you inject a sine wave? It may be almost ok at very low output but then starting to get severely distorted as you increase amplitude. Is this what you observe?
Tell us how these tests go in order to continue helping you.
I have the strange and tiring sensation that this will trigger and endless discussion and will hijack the purpose if this thread, to help Rob. But, anyway, I have to say that I have got very clean and useful readings of upper mosfet Vgs and bootstrap voltage quality measuring this way, of course this has to be done carefully and isolating the oscilloscope in a controlled lab environment, and ideally using a HF probe, if available.
That's why I have shared my experience, I think that it will help Rob determine the real cause of his problem.
That's why I have shared my experience, I think that it will help Rob determine the real cause of his problem.
I don't like that kind of discussions.
What I mean is that VB and VS nodes obviously experience very high dV/dt transients. Any long wire connected to these nodes can radiate enough to disturb a modulator built on breadboard or any equpment around. Also, the load added when connecting something to these nodes can boost ringing and disturb the power stage.
A multimeter is a good antenna too.
What I have done sometimes is to use two oscilloscope probes and connect the tip of one to VS and the tip of the other to VB or to the gate signal. My oscilloscope can show the voltage difference between both channels, thus acting as a poor man's differential probe.
For example, my current project (200V 42A full bridge) is free of ringing, everything is carefully damped, and crossover time is 35ns, it's not that fast. But when I connect one oscilloscope probe to the switching node, I can see some ~90Mhz ringing appear on the gate signal monitored with the other probe. Even the apparently harmless load of the probe is already causing slight disturbance on the circuit, or the oscilloscope, or both!
In other words, be very careful when connecting things to switching nodes. Ground measuring equipment only to "quiet" nodes.
What I mean is that VB and VS nodes obviously experience very high dV/dt transients. Any long wire connected to these nodes can radiate enough to disturb a modulator built on breadboard or any equpment around. Also, the load added when connecting something to these nodes can boost ringing and disturb the power stage.
A multimeter is a good antenna too.
What I have done sometimes is to use two oscilloscope probes and connect the tip of one to VS and the tip of the other to VB or to the gate signal. My oscilloscope can show the voltage difference between both channels, thus acting as a poor man's differential probe.
For example, my current project (200V 42A full bridge) is free of ringing, everything is carefully damped, and crossover time is 35ns, it's not that fast. But when I connect one oscilloscope probe to the switching node, I can see some ~90Mhz ringing appear on the gate signal monitored with the other probe. Even the apparently harmless load of the probe is already causing slight disturbance on the circuit, or the oscilloscope, or both!
In other words, be very careful when connecting things to switching nodes. Ground measuring equipment only to "quiet" nodes.
Yes, as I said, you may get somewhat distorted or false readings when measuring this way, but may experience is that you can get good waveforms even when the leads can form some antenna and of course produce lots of radiating EMI in the lab. But depending on what you need to measure, it may be more than ok.
The differential measuring technique works well with analog scopes, but not so well with digital ones, specially when the waveforms you need to capture are much larger than their difference, then the resolution of the scope (usually 8 bits) limits what you get and perhaps you are only 3 or 4 bits left for the measure you are interested in, and that's very little.
In this case, where the rails are not high (25V) and the HO or VB voltages are in the same order of magnitude, it may work well, anyway.
What we agree, Rob, is that you need to measure Vb-Vs and HO-Vs to see what's happening. Please tell us what you get.
The differential measuring technique works well with analog scopes, but not so well with digital ones, specially when the waveforms you need to capture are much larger than their difference, then the resolution of the scope (usually 8 bits) limits what you get and perhaps you are only 3 or 4 bits left for the measure you are interested in, and that's very little.
In this case, where the rails are not high (25V) and the HO or VB voltages are in the same order of magnitude, it may work well, anyway.
What we agree, Rob, is that you need to measure Vb-Vs and HO-Vs to see what's happening. Please tell us what you get.
sorry i have taken so long to reply, i am just so busy with other projects. ok so i sat down with the amp today with the little time i had and took the following measurements:
Conditions:
Positive rail voltage: 20V
Negative rail voltage: -20V
No load
MOSFETS used: IRFB4019PbF
Measurements:
VAA(pin1) with respect to GND(pin2) = 7.3V
VAA(pin1) with respect to VSS(pin6) = 13.1V
VSS(pin6) with respect to GND(pin2) = -5.7V
VAA(pin1) with respect to COM(pin10) = 26.2V
VSS(pin6) with respect to COM(pin10) = 13.2V
VB(pin15) with respect to COM(pin10) = 38.9V
VS(pin13) with respect to COM(pin10) = 38.6V
CSD(pin5) with respect to GND(pin2) = 7.3V
Vref(pin7) with respect to GND(pin2) = -19.0V
Ocset(pin8) with respect to GND(pin2) = -19.0V
DT(pin9) with respect to GND(pin2) = -19.0V
Vcc(pin12) with respect to GND(pin2) = -19.0V
VS(pin13) with respect to GND(pin2) = 20V
VB(pin15) with respect to GND(pin2) = 20V
CSH(pin16) with respect to GND(pin2) = 20V
I have also attached some pictures of the breadboard layout and also a copy of the typical application circuit that i am basing my amp on. I am confused regarding the grounding and the different grounding potentials that exist in this design.
Conditions:
Positive rail voltage: 20V
Negative rail voltage: -20V
No load
MOSFETS used: IRFB4019PbF
Measurements:
VAA(pin1) with respect to GND(pin2) = 7.3V
VAA(pin1) with respect to VSS(pin6) = 13.1V
VSS(pin6) with respect to GND(pin2) = -5.7V
VAA(pin1) with respect to COM(pin10) = 26.2V
VSS(pin6) with respect to COM(pin10) = 13.2V
VB(pin15) with respect to COM(pin10) = 38.9V
VS(pin13) with respect to COM(pin10) = 38.6V
CSD(pin5) with respect to GND(pin2) = 7.3V
Vref(pin7) with respect to GND(pin2) = -19.0V
Ocset(pin8) with respect to GND(pin2) = -19.0V
DT(pin9) with respect to GND(pin2) = -19.0V
Vcc(pin12) with respect to GND(pin2) = -19.0V
VS(pin13) with respect to GND(pin2) = 20V
VB(pin15) with respect to GND(pin2) = 20V
CSH(pin16) with respect to GND(pin2) = 20V
I have also attached some pictures of the breadboard layout and also a copy of the typical application circuit that i am basing my amp on. I am confused regarding the grounding and the different grounding potentials that exist in this design.
Hello,
You are not likely going to get decent results with that layout, but, anyway, at least the DC measurements should be ok, and they are not.
For a start, your VB is not ok, you are getting 38V with respect to COM, when the IRS2092 has an internal 20V zener, so something is not connected properly around this point. It is not that the amplifier has not started or something and you don't get the proper Vb, it is that it doesn't seem to be connected to the proper IC pin.
Also, your output is stick to the positive rail. Please try with some load connected just in case it hasn't started oscillating, but to be sincere, I would not spend much more time until you fix the VB issue, because it is clearly indicating that you have some major errors.
Once this is ok, you WILL need an oscilloscope. Don't even think on debugging and putting it to work with only the measurements of a voltmeter.
You also say you don't understand the different references, and that's the most important point to have clear in a Class-D design, so please spend some time reading the datasheets throughly or ask here. Basically, you have a GND reference at the input of the chip, and the output is referenced to the negative rail (COM pin). Also, the VB is referenced to the switching pin (VS) that should swing from -Vbus to +Vbus.
You are not likely going to get decent results with that layout, but, anyway, at least the DC measurements should be ok, and they are not.
For a start, your VB is not ok, you are getting 38V with respect to COM, when the IRS2092 has an internal 20V zener, so something is not connected properly around this point. It is not that the amplifier has not started or something and you don't get the proper Vb, it is that it doesn't seem to be connected to the proper IC pin.
Also, your output is stick to the positive rail. Please try with some load connected just in case it hasn't started oscillating, but to be sincere, I would not spend much more time until you fix the VB issue, because it is clearly indicating that you have some major errors.
Once this is ok, you WILL need an oscilloscope. Don't even think on debugging and putting it to work with only the measurements of a voltmeter.
You also say you don't understand the different references, and that's the most important point to have clear in a Class-D design, so please spend some time reading the datasheets throughly or ask here. Basically, you have a GND reference at the input of the chip, and the output is referenced to the negative rail (COM pin). Also, the VB is referenced to the switching pin (VS) that should swing from -Vbus to +Vbus.
Thanks very much for the comments. going to investigate the situtaion with VB now. regarding the layout is there any recconmendations you would make. I know it is not really ideal to build any type of high frequency circuit on a breadboard. I would be looking to implement this design/layout on to a PCB.
i have done a slight re-design. i am using external Positive and Negative regulators now for VAA and VSS. I also tried to shorten the length of all wires, legs of caps & resistors as short as possible while still retaining the appropriate spacing between the various components. i would like to to keep the power output stage (MOSFET's) disconnected from the IRS2092 chip completely (for the time being) and try to get some decent initial conditions with the IRS2092. Is this plausable? Once i am happy with the measurements and waveforms at various points of the IRS2092 i will then connect the power output stage and build upon that. what waveforms should i expect to see on pins HO (pin 14) & LO (pin 11)? square waveforms? of what amplitude ?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.