Hi I got around 900Wrms with two irs2092 + irfb4227
driving a 4ohm load in BTL with +/-55V rails.
Now I am designing an amp with irs2092 with +/-85V rails.
It should in theory be able to drive an 8ohm load with at least 300Wrms. I think that if you bridge it and if the power supply keeps up you will be at at least 1200-1300Wrms on 8ohm
@moer
if there is a current limit protection for me it is a nonsense to disable it or to set it to a trip level much higher that required.
If you have the OCP tipping is probably because you have too much inductance between the lower side source and the the -B power supply. You need to put a plane on the PCB in order to reduce that inductance. -B is the power reference ground for the IRS2092 (the chip substrate is connected to that pin).
This inductance probably creates some spikes that makes the current limit trip before than expected.
@ponymxy
Look at the IRS20955 datasheet, there is an explanation of how to set the current limit properly. In the iraudamp 5 reference design the explanation is not clear at all and in the irs2092 datasheet there is no explanation.
ciao
-marco
driving a 4ohm load in BTL with +/-55V rails.
Now I am designing an amp with irs2092 with +/-85V rails.
It should in theory be able to drive an 8ohm load with at least 300Wrms. I think that if you bridge it and if the power supply keeps up you will be at at least 1200-1300Wrms on 8ohm
@moer
if there is a current limit protection for me it is a nonsense to disable it or to set it to a trip level much higher that required.
If you have the OCP tipping is probably because you have too much inductance between the lower side source and the the -B power supply. You need to put a plane on the PCB in order to reduce that inductance. -B is the power reference ground for the IRS2092 (the chip substrate is connected to that pin).
This inductance probably creates some spikes that makes the current limit trip before than expected.
@ponymxy
Look at the IRS20955 datasheet, there is an explanation of how to set the current limit properly. In the iraudamp 5 reference design the explanation is not clear at all and in the irs2092 datasheet there is no explanation.
ciao
-marco
IR class D chips
I have planes on our PCBs and by IR's own admission the high and low side current trip ccts are overly sensitive.
The PDF which they sent me the other day for their "1Kw" version demands the use of totem poles when using rails as high as +/-85v.
PS I have looked at the data sheet of the 20955 for the past 2 years and know how to calculate the over current trip.
Again by Jun Honda's (The guy who designed this chip) own admission, the current trip point cannot be set below 20-22 amps as the system will shut down all the time. The formulas supplied allow this level of OC trip points but it does not work.
I am using IRFP4227 in our 600w amplifier but will change to 3+3 of IRFB4020 each with their own totem driver.
Run the rails at +/-85v with no totem poles and the chip WILL burn up.
I have planes on our PCBs and by IR's own admission the high and low side current trip ccts are overly sensitive.
The PDF which they sent me the other day for their "1Kw" version demands the use of totem poles when using rails as high as +/-85v.
PS I have looked at the data sheet of the 20955 for the past 2 years and know how to calculate the over current trip.
Again by Jun Honda's (The guy who designed this chip) own admission, the current trip point cannot be set below 20-22 amps as the system will shut down all the time. The formulas supplied allow this level of OC trip points but it does not work.
I am using IRFP4227 in our 600w amplifier but will change to 3+3 of IRFB4020 each with their own totem driver.
Run the rails at +/-85v with no totem poles and the chip WILL burn up.
500w amp using IRS20955
Dear Raidfibre,
If the 500w is rated at 4 ohms, you can run +/-85v with Totem drivers between the 20955 and the MOSFETS. I use 3+3 of IRFB4020 for our 600w 4 ohm amplifiers but these are supplied from a well regulated SMPS with PFC.
I disable the IRS20955 OC protection as it must be set at well over 80 amps to work (This is useless anyway)
I use a current transformer in my speaker circuit (It is there anyway as the pro amps we make have microprocessor control and the speaker current is one of the items which the microprocessor reports on).
it is a T50-26 from Micrometals with 100 turns on the secondary and 1 turn for the primary. I shape the output from the 100t secondary to give me a flat response from 50Hz to 10KHz, then I full wave rectify this to give me a DC voltage proportional to the speaker current. Using the value of this DC signal, throw it into a comparator, and use this output to shut the CSD pin down
Regards
Stephen Mantz
Zed Audio Corporation
Dear Raidfibre,
If the 500w is rated at 4 ohms, you can run +/-85v with Totem drivers between the 20955 and the MOSFETS. I use 3+3 of IRFB4020 for our 600w 4 ohm amplifiers but these are supplied from a well regulated SMPS with PFC.
I disable the IRS20955 OC protection as it must be set at well over 80 amps to work (This is useless anyway)
I use a current transformer in my speaker circuit (It is there anyway as the pro amps we make have microprocessor control and the speaker current is one of the items which the microprocessor reports on).
it is a T50-26 from Micrometals with 100 turns on the secondary and 1 turn for the primary. I shape the output from the 100t secondary to give me a flat response from 50Hz to 10KHz, then I full wave rectify this to give me a DC voltage proportional to the speaker current. Using the value of this DC signal, throw it into a comparator, and use this output to shut the CSD pin down
Regards
Stephen Mantz
Zed Audio Corporation
IRS20955
I'm confused as to why you would need to set the OCP to over 80 amps. Is this because it is too sensitive?
Yes and any spikes from the MOSFET switching action aggrivates the situation. I have been going on with this issue with International Rectifier for almost 3 years and the problem is quite simple. When they monitor the RDS on of the upper and lower MOSFETs, they do this internally and there is no provision to incorporate a low pass filter into the sensing leads.
I also use a discrete circuit I designed and I monitor the RDs on BUT I use LP filters into my comparators. No problem
The one advantge of the IR part is the size and the easily set dead time
Stephen Mantz
Zed Audio Corp.
I'm confused as to why you would need to set the OCP to over 80 amps. Is this because it is too sensitive?
Yes and any spikes from the MOSFET switching action aggrivates the situation. I have been going on with this issue with International Rectifier for almost 3 years and the problem is quite simple. When they monitor the RDS on of the upper and lower MOSFETs, they do this internally and there is no provision to incorporate a low pass filter into the sensing leads.
I also use a discrete circuit I designed and I monitor the RDs on BUT I use LP filters into my comparators. No problem
The one advantge of the IR part is the size and the easily set dead time
Stephen Mantz
Zed Audio Corp.
Guys, forget about 20955, as you already forgot about 20954, this stuff is really unstable and now officially isn't recommended to use any more, only 20957 is ok. 
http://www.irf.com/technical-info/appnotes/an-1141.pdf
http://www.irf.com/technical-info/appnotes/an-1141.pdf
IRS20955
Guys, forget about 20955, as you already forgot about 20954, this stuff is really unstable and now officially isn't recommended to use any more, only 20957 is ok.
Spoke to IR here in El segundo CA and the 20957 is the SAME chip as the 20955 BUT with longer dead time
Stephen Mantz
Zed Audio Corp.
CA USA
Guys, forget about 20955, as you already forgot about 20954, this stuff is really unstable and now officially isn't recommended to use any more, only 20957 is ok.
Spoke to IR here in El segundo CA and the 20957 is the SAME chip as the 20955 BUT with longer dead time
Stephen Mantz
Zed Audio Corp.
CA USA
Re: IRS20855 Unstable
Basically what IVX said then, all along. You might like to read his link again. As it states the difference is that it ignores pulses less than 50nS =! dead time. It also mentions a shorter propagation delay, and a longer deadtime window. Only on DT4 setting is deadtime longer, at 80nS instead of 45nS typically, DT1-3 is unchanged with one condition.
AN-1141 that Ivan linked to, under Upgrading IRS20955 with IRS20957, mentions it's a direct upgrade with longer deadtime for DT4, while for DT1-3 they don't recommend VCC greater than 15V, where no doubt it falls out of spec.
Maybe it just needs a "totem" and ideal mosfet to fix the OC trip level.. but it is funny that nothing of that changed at all, even with your close friend at IR stating through you that it was a known problem. I guess then this chip should be avoided as well.
MOER said:
Basically what IVX said then, all along. You might like to read his link again. As it states the difference is that it ignores pulses less than 50nS =! dead time. It also mentions a shorter propagation delay, and a longer deadtime window. Only on DT4 setting is deadtime longer, at 80nS instead of 45nS typically, DT1-3 is unchanged with one condition.
AN-1141 that Ivan linked to, under Upgrading IRS20955 with IRS20957, mentions it's a direct upgrade with longer deadtime for DT4, while for DT1-3 they don't recommend VCC greater than 15V, where no doubt it falls out of spec.
Maybe it just needs a "totem" and ideal mosfet to fix the OC trip level.. but it is funny that nothing of that changed at all, even with your close friend at IR stating through you that it was a known problem. I guess then this chip should be avoided as well.
IR20955/2092S
Dear Classdphile,
I have had many meetings and phone calls with Jun Honda who is the prime designer behind these chips.
Since these chips were designed to be used with their direct FETS only, the layout using these tiny parts can really be optimized. Read the PDF of the appliciation board (which I have) and IR set the OC to 30A on both the high adn low sides. The MOSFETS are rated at about 6A.
This is not a good thing. A 6A MOSFET with a 30A over current threshold.
We use these chips in professional amplifiers and fortunately we have full microprocessor control and one of the functions is that we have to monitor the speaker current and voltage. So I use a small current xfr and develop a DC control signal which is fed to the microprocessor. I use this DC control signal (Proportional to the speaker current) and throw it into a comparator whose output shuts of the CSD pin of the IR chip.
I set the OC thresholds on the chip to well over 100amps so that they are effectively out of circuit.
Any tiny glitvhes on the supply rails and the IR chip responds simply by virctue of the idling current in the output stage
Stephen Mantz
Zed Audio Corp.
Dear Classdphile,
I have had many meetings and phone calls with Jun Honda who is the prime designer behind these chips.
Since these chips were designed to be used with their direct FETS only, the layout using these tiny parts can really be optimized. Read the PDF of the appliciation board (which I have) and IR set the OC to 30A on both the high adn low sides. The MOSFETS are rated at about 6A.
This is not a good thing. A 6A MOSFET with a 30A over current threshold.
We use these chips in professional amplifiers and fortunately we have full microprocessor control and one of the functions is that we have to monitor the speaker current and voltage. So I use a small current xfr and develop a DC control signal which is fed to the microprocessor. I use this DC control signal (Proportional to the speaker current) and throw it into a comparator whose output shuts of the CSD pin of the IR chip.
I set the OC thresholds on the chip to well over 100amps so that they are effectively out of circuit.
Any tiny glitvhes on the supply rails and the IR chip responds simply by virctue of the idling current in the output stage
Stephen Mantz
Zed Audio Corp.
@ Mag
I have been trying to design a very efficient, very tough, class B amp. If you read the books of Douglas Self, he is very disparaging about quasi-complimentary MOSFET amplifiers. I feel I have no choice but to use that topology because P channels are becoming rare and expensive. It struck me I should go to class D, like yourself I have no need for high frequencies so I thought I could risk a simple layout.
My question to you is did you have the manufacturers information for PCB layout, if not, did you just take a risk? I suppose the question is if so, did you have any problems?
Sorry to ask this, you may have already answered this question, I'm doing this long distance with my son typing it over the computer, his knowledge of electronics isn't quite enough to answer everything. He has glanced over your replies. I have set him the task today of researching the IRS2092 chip which I discovered this morning. From what he describes, it's a very good project and I can see no reason to look at a class B design. To put you in the picture I would run the class B design continuously at near clipping. I would run from switch mode power supplies so as to define the clipping point. I would be running at full power most of the time, I would expect an efficiency on the class B of about 70% if very lucky. The efficiency you get would be a considerable advantage.
Many thanks if you have time to reply,
Nigel.
I have been trying to design a very efficient, very tough, class B amp. If you read the books of Douglas Self, he is very disparaging about quasi-complimentary MOSFET amplifiers. I feel I have no choice but to use that topology because P channels are becoming rare and expensive. It struck me I should go to class D, like yourself I have no need for high frequencies so I thought I could risk a simple layout.
My question to you is did you have the manufacturers information for PCB layout, if not, did you just take a risk? I suppose the question is if so, did you have any problems?
Sorry to ask this, you may have already answered this question, I'm doing this long distance with my son typing it over the computer, his knowledge of electronics isn't quite enough to answer everything. He has glanced over your replies. I have set him the task today of researching the IRS2092 chip which I discovered this morning. From what he describes, it's a very good project and I can see no reason to look at a class B design. To put you in the picture I would run the class B design continuously at near clipping. I would run from switch mode power supplies so as to define the clipping point. I would be running at full power most of the time, I would expect an efficiency on the class B of about 70% if very lucky. The efficiency you get would be a considerable advantage.
Many thanks if you have time to reply,
Nigel.
Class D amplifier
Dear Nigel,
Depending on waht power output you require and into waht impedance you intend to drive the amplifier, one must be carefull with these IRS20955 and IRS2092 chips.
They are not happy when driven into clipping and I would advise a compressor prior to the class D as I do this and have ZERO failures of the class D.
The "flyback" diodes after the filter to each rail MUST use high speed high current diodes only. Do not use 1 amp diodes.
If your requirement is for lower frequencies by all means lower the switching frequency but not below 200KHz as you will incurr losses in the output coil.
Make sure you use MOSFETS which are suitably rated.
PCB layout is critical and fortunately for me I do have lots of experience with this.
SMD parts are a 100% must in these designs.
Stephen
Dear Nigel,
Depending on waht power output you require and into waht impedance you intend to drive the amplifier, one must be carefull with these IRS20955 and IRS2092 chips.
They are not happy when driven into clipping and I would advise a compressor prior to the class D as I do this and have ZERO failures of the class D.
The "flyback" diodes after the filter to each rail MUST use high speed high current diodes only. Do not use 1 amp diodes.
If your requirement is for lower frequencies by all means lower the switching frequency but not below 200KHz as you will incurr losses in the output coil.
Make sure you use MOSFETS which are suitably rated.
PCB layout is critical and fortunately for me I do have lots of experience with this.
SMD parts are a 100% must in these designs.
Stephen
Hello Moer,im trying to activate the internal protection in a safe way,but i have to adjust it to a very high level to work properly.I had changed the lay out ,as ir note described,it help but its not working as it could be,do think its possible to use oc protection if we use a good rc snubber at switching node?Maybe im not tunning it very well.Fly back diodes at switching node too?Why dont you use 1a fd as ir suggested?
The worse,my 20 mhz scope cant catch any spike image at square wave even at high power ,maybe a lot of megahertz up..
The worse,my 20 mhz scope cant catch any spike image at square wave even at high power ,maybe a lot of megahertz up..
During my first baby steps with class D I was also struggling with undesired shut downs of the IRS20954.
Later I learned that I had quite heavy resonances in my PCB. The resonances were in the range of 30MHz. At that time I was already using a concept without the IR chip and I did never try again how the IR-Chip would perform with a proper PCB and snubbered MosFets. But I could imagine that many of the reported issues with "oversensitive" shut downs could be related to short comings of the PCB design and missing or ineffective snubbers. Typically such resonances are ranging between 20MHz...100MHz and become especially visible in hard switching conditions - often not during idle.
Even if the shut down functions of the IR chips are not really oversensitive, but just fast, then such resonances might probably trigger them.
In fact I could imagine that IR has changed their shut down functions in order to become more tolerant to some imperfection of the application.
Later I learned that I had quite heavy resonances in my PCB. The resonances were in the range of 30MHz. At that time I was already using a concept without the IR chip and I did never try again how the IR-Chip would perform with a proper PCB and snubbered MosFets. But I could imagine that many of the reported issues with "oversensitive" shut downs could be related to short comings of the PCB design and missing or ineffective snubbers. Typically such resonances are ranging between 20MHz...100MHz and become especially visible in hard switching conditions - often not during idle.
Even if the shut down functions of the IR chips are not really oversensitive, but just fast, then such resonances might probably trigger them.
In fact I could imagine that IR has changed their shut down functions in order to become more tolerant to some imperfection of the application.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.