Sorry Subwo1 I should have stated the case more clearly.
The topology that is giving me headaches is a CT Push pull converter not a half or full bridge. I have been doing some reading and many people claim that the CT Push Pull is troublesome because of symmetry issues.
I would like to know though if I could use a Full Bridge for a converter that is powered by a car battery. My problem is that I have 2 x 276Wrms amplifier modules that require + and - 63 V at 3.76A per side (1 module) x 2 modules. If I test the converter on 1 module only the output voltages hold up very well. The moment I introduce the second amp the output voltage falls to about + and - 54V and thus I am only able to get out around 200Wrms/ch into a 4R load.
I am using 6 Mosfets per side and full wave rectifier for the output consisting of Fast recovery TO220 diodes.
Any help or directions is most appreciated.
The topology that is giving me headaches is a CT Push pull converter not a half or full bridge. I have been doing some reading and many people claim that the CT Push Pull is troublesome because of symmetry issues.
I would like to know though if I could use a Full Bridge for a converter that is powered by a car battery. My problem is that I have 2 x 276Wrms amplifier modules that require + and - 63 V at 3.76A per side (1 module) x 2 modules. If I test the converter on 1 module only the output voltages hold up very well. The moment I introduce the second amp the output voltage falls to about + and - 54V and thus I am only able to get out around 200Wrms/ch into a 4R load.
I am using 6 Mosfets per side and full wave rectifier for the output consisting of Fast recovery TO220 diodes.
Any help or directions is most appreciated.
Archimedes, here is an idea. On the push-pull circuit, it is possible that your transformer may be starting to saturate. More transformer core volume may help. If you have a scope, look for a sharp rise in MOSFET current at the end of the duty cycle.
It may help to raise the frequency some if you are getting saturation. Then, you will probably have to reduce the number of turns on both the primary and secondary. ?
It may help to raise the frequency some if you are getting saturation. Then, you will probably have to reduce the number of turns on both the primary and secondary. ?
Hello Archimedes,
It is very difficult to diagnose a problem without any real data on what is going on.
There have been some good recommendations so far. It is possible that your core is the wrong size for the frequency and power level you want to use.
It is also possible that your windings and source have too high resistance.
What do you have for capacitors on the center tap of the primary side?
What frequency are you running at?
Do you have any pictures or schematics to show us?
It is very difficult to diagnose a problem without any real data on what is going on.
There have been some good recommendations so far. It is possible that your core is the wrong size for the frequency and power level you want to use.
It is also possible that your windings and source have too high resistance.
What do you have for capacitors on the center tap of the primary side?
What frequency are you running at?
Do you have any pictures or schematics to show us?
Thanks to the moderator for moving this.
The trafo is a Siemens EPCOS toroid R58,3 × 40,8 × 17,6 (Ae=152.4 sq.mm, Ve=23230 cu.mm) with N87 material. It has Pri:2+2 (8 x 1.61mm wire per primary)
and Sec:14+14 (40 x 0.28 wires twisted per secondary) running at 43kHz. There are 5 (not 6 as previously mentioned) IRFZ44's per switching side. These are driven by a complimentary follower pair made from BD139 and BD140 per side. The chip is an SG3525. I have 4 x 4700uF caps near the centre tap. There is a 22uH choke on the input from the battery + to the caps +.
When I tested the psu I noticed that it had pulled almost 110A at 10.6V battery input measured with a clamp.
The voltages on the secondary of the converter was then + and - 54V at +/- 3A per side. Looking at the waveform on a scope the pk.pk value measured at any drain had fallen to 21V and the waveform was completely square (Full on).
Is the transformer to small ? Just curious Has anyone ever built a Full bridge converter powered by a car battery. Is it really impractical to do so ?
Phew... this looks like a thesis
The trafo is a Siemens EPCOS toroid R58,3 × 40,8 × 17,6 (Ae=152.4 sq.mm, Ve=23230 cu.mm) with N87 material. It has Pri:2+2 (8 x 1.61mm wire per primary)
and Sec:14+14 (40 x 0.28 wires twisted per secondary) running at 43kHz. There are 5 (not 6 as previously mentioned) IRFZ44's per switching side. These are driven by a complimentary follower pair made from BD139 and BD140 per side. The chip is an SG3525. I have 4 x 4700uF caps near the centre tap. There is a 22uH choke on the input from the battery + to the caps +.
When I tested the psu I noticed that it had pulled almost 110A at 10.6V battery input measured with a clamp.
The voltages on the secondary of the converter was then + and - 54V at +/- 3A per side. Looking at the waveform on a scope the pk.pk value measured at any drain had fallen to 21V and the waveform was completely square (Full on).
Is the transformer to small ? Just curious Has anyone ever built a Full bridge converter powered by a car battery. Is it really impractical to do so ?
Phew... this looks like a thesis
If you post the following information, it may be helpful in troubleshooting the supply.
Core data (link to datasheet showing core material and dimensions would be very helpful).
Number of primary turns (for example, 4+4 or 5+5).
Number of secondary turns.
The amount of copper on the primary and secondary (number of conductors and size of conductors for each turn).
The operating frequency.
The part numbers of the FETs and the rectifier diodes.
A schematic.
Core data (link to datasheet showing core material and dimensions would be very helpful).
Number of primary turns (for example, 4+4 or 5+5).
Number of secondary turns.
The amount of copper on the primary and secondary (number of conductors and size of conductors for each turn).
The operating frequency.
The part numbers of the FETs and the rectifier diodes.
A schematic.
I think you are driving your transformer with too high flux density:
From your data (2+2 primary, 43KHz, 1.52cm^2 Ac), and 12V input, your peak flux density is about 2300 gauss, perhaps too high. That could lead to transformer saturation.
Try to increase frequency to 75-80KHz or increase primary turns number to something like 3+3 or 4+4 turns.
(If you increase frequency you will have greater mosfet losses than a converter at 43KHz working properly, but in your case, if the transformer is saturating, they must be suffering a lot just now! even at that low freq.)
Your converter should work well with a moderate load before going to high loads. Try with 40-50W first.
Apart from this, I would like to hear why is so uncommon (in fact I have seen none) to find a half-bridge converter in car audio, all of them seem to be push-pull. Is it only a matter of complexity or cost of the gate drive circuit?
Hope this helps.
From your data (2+2 primary, 43KHz, 1.52cm^2 Ac), and 12V input, your peak flux density is about 2300 gauss, perhaps too high. That could lead to transformer saturation.
Try to increase frequency to 75-80KHz or increase primary turns number to something like 3+3 or 4+4 turns.
(If you increase frequency you will have greater mosfet losses than a converter at 43KHz working properly, but in your case, if the transformer is saturating, they must be suffering a lot just now! even at that low freq.)
Your converter should work well with a moderate load before going to high loads. Try with 40-50W first.
Apart from this, I would like to hear why is so uncommon (in fact I have seen none) to find a half-bridge converter in car audio, all of them seem to be push-pull. Is it only a matter of complexity or cost of the gate drive circuit?
Hope this helps.
There is nothing inherently wrong with the push-pull topology, and full bridges are exactly as prone to tranformer saturation as push-pulls if things are not done properly. At least 500ns of dead time should be provided, and proper RC snubbers are required in the primary side in order to allow for that small amount of unavoidable gate drive asymmetry, that is always present, to be automatically compensated by faster or slower rise times, depending on the own primary magnetizing current, ie: the longer cycle ends with higher magnetizing current, thus the snubber capacitor is charged faster compensating flux imbalance).
Also, the voltage drop that you describe is absolutely normal for non-regulated converters, it happens because the leakage inductance of the transformer starts to dominate when some current level is reached, and more and more voltage drop is required across that inductance in order to reach the required transformer current level within each cycle. There are several ways to circumvent that: The operating frecuency may be reduced (thus making cycles longer and providing more time for the leakage inductance to be energyzed) if you core allows for that, or more SMPS transformers (or entire cells with switched and diodes) may be paralelled in order to reduce the impact of leakage inductance, or a regulated topology may be considered (altough that would require higher peak currents to be handled, a lot of low ESR filter capacitors and probably a bulky coupled output inductor).
Note that when using multiple transformers, you can connect the secondaries in series, thus reducing the number of turns required in each one and making winding task easier.
Also, are you sure that your turns ratio is 2:14? That should be providing more than +-80V when idle. Is your test battery in good condition? You sould be getting 11,5V or more with 100A load. Finally, don't trust amperimetric clamps at all, they are not likely to cope well with the pulsed current consumption produced by amplifiers and SMPS, they are usually intended for DC measurements only. Use an oscilloscope instead, so that you can see actual waveforms and peak values.
Pierre: A half bridge is almost the same as a push pull, but replacing the coupling capacitor by a dual transformer winding. Then again, where are you going to find capacitors rated at several thousands microfarads and a hundred amperes of ripple current?
Also, the voltage drop that you describe is absolutely normal for non-regulated converters, it happens because the leakage inductance of the transformer starts to dominate when some current level is reached, and more and more voltage drop is required across that inductance in order to reach the required transformer current level within each cycle. There are several ways to circumvent that: The operating frecuency may be reduced (thus making cycles longer and providing more time for the leakage inductance to be energyzed) if you core allows for that, or more SMPS transformers (or entire cells with switched and diodes) may be paralelled in order to reduce the impact of leakage inductance, or a regulated topology may be considered (altough that would require higher peak currents to be handled, a lot of low ESR filter capacitors and probably a bulky coupled output inductor).
Note that when using multiple transformers, you can connect the secondaries in series, thus reducing the number of turns required in each one and making winding task easier.
Also, are you sure that your turns ratio is 2:14? That should be providing more than +-80V when idle. Is your test battery in good condition? You sould be getting 11,5V or more with 100A load. Finally, don't trust amperimetric clamps at all, they are not likely to cope well with the pulsed current consumption produced by amplifiers and SMPS, they are usually intended for DC measurements only. Use an oscilloscope instead, so that you can see actual waveforms and peak values.
Pierre: A half bridge is almost the same as a push pull, but replacing the coupling capacitor by a dual transformer winding. Then again, where are you going to find capacitors rated at several thousands microfarads and a hundred amperes of ripple current?
You mentioned voltage feedback so it sounds like a regulated supply. The combination of the input voltage sagging and the regulator hitting the max duty cycle will lead to a significant/sharp drop in output.
As was mentioned, the transformer may be saturating. Many times, you can hear the transformer making noise when it saturates. This is most noticeable when playing music. Also, when it saturates, you may see a flat line at ~1/2 B+ when looking at the waveform at a low sweep rate (where hundreds of cycles are displayed at once).
From my experience, you'll get the best overall performance at ~30kHz and 4+4 on the primary.
As was mentioned, the transformer may be saturating. Many times, you can hear the transformer making noise when it saturates. This is most noticeable when playing music. Also, when it saturates, you may see a flat line at ~1/2 B+ when looking at the waveform at a low sweep rate (where hundreds of cycles are displayed at once).
From my experience, you'll get the best overall performance at ~30kHz and 4+4 on the primary.
Current-Mode -v- Voltage-Mode
Archimedes,
Voltage-mode controllers like the '3525 cannot adequately deal with the flux imbalance that can occur with C.T. push-pull topology.
Current-mode is needed, ans this can be achieved with such chips like the SG1846/2846/3846, UC3825, or the MC33025. The first one is a standard-speed chip, while the last two are high speed (2MHz oscillator) chips.
I have used the 33025 before with good results. See my other threads where I talk about this chip. It could solve your problems
Steve
Archimedes,
Voltage-mode controllers like the '3525 cannot adequately deal with the flux imbalance that can occur with C.T. push-pull topology.
Current-mode is needed, ans this can be achieved with such chips like the SG1846/2846/3846, UC3825, or the MC33025. The first one is a standard-speed chip, while the last two are high speed (2MHz oscillator) chips.
I have used the 33025 before with good results. See my other threads where I talk about this chip. It could solve your problems
Steve
Thanks, Eva. Sorry for going a bit off-topic. Although your explanations about why half-bridge is not used seems to go in the right direction, there is still something I don't understand:
-In half bridge, the usual thing is to make a Vcc/2 voltage by means of large capacitors and couple the transformer to that point. But I have also seen single filter capacitors (as in push-pull), the transformer being coupled to GND (or to the mosfets middle) by means of a _small_ capacitor. Is that small-valued capacitor (say, 10uF) with high current rating so difficult to find, or the problem are the big ones? In that case, with push-pull it is the same, isn't it?
Thanks for your help!
-In half bridge, the usual thing is to make a Vcc/2 voltage by means of large capacitors and couple the transformer to that point. But I have also seen single filter capacitors (as in push-pull), the transformer being coupled to GND (or to the mosfets middle) by means of a _small_ capacitor. Is that small-valued capacitor (say, 10uF) with high current rating so difficult to find, or the problem are the big ones? In that case, with push-pull it is the same, isn't it?
Thanks for your help!
Keeping frequency constant, the required value for that "small" capacitor is inversely proportional to the input voltage squared, and proportional to the output power.
For a 300V 200W half bridge 1uF is routinely used, so figure out how much capacitance is required for a 12V 1KW half bridge. Also, such a half bridge would put only 6V to the primary winding, so current would get doubled (6250uF with 166A ripple)...
For a 300V 200W half bridge 1uF is routinely used, so figure out how much capacitance is required for a 12V 1KW half bridge. Also, such a half bridge would put only 6V to the primary winding, so current would get doubled (6250uF with 166A ripple)...
Perry I think you may have a point there. When the output voltage falls off I can hear a slight whine from the transformer. I think I shall try the suggestion with 4+4 on the pri @30kHz. How much of copper do you think is practical ?
Eva I have never seen any car amplifier PSU's incorporating current mode control. How would something like this be done practically anyway. I think the SG3526 does allow current mode control. The turns ratio is 1:7. When I removed the voltage feedback at idle I noticed the voltage shoot up to +/- 92V at 12.6V input. The battery is fully charged. The battery voltage only falls to 10.5 (measured at the CT of trafo) when I think the trafo saturates. Without saturation it stays at 11.2V. (65AH battery).
How would you use a scope to measure current with ordinary probes ?
Eva I was referring to a Full bridge not the half bridge. I know that you will need twice the amount of switches but since IRFZ44's cost here around 70 US cents each it does not seem that unfeasible. Furthermore the way I look at it is that you only need 12 x 1.61mm wires for the Pri and not 24. This should greatly simply the trafo winding process. What do you think?
Can anyone please tell me how much inductance is necessary for the input and output chokes ? I chose values that I saw commonly used. Is there formulae for this purpose so that I don't have to do thumb sucking ?
Can I please ask the gurus out there to take a look at the attached spreadsheet and give me any feedback if possible.
I use it whenever I wind trafo's.
PS. Do you think it will be more feasible to wind a 305.9mm Ac (I have this too) or will the 152.4 Ac give me the required 1000W ?
thanks everyone for your input!
Eva I have never seen any car amplifier PSU's incorporating current mode control. How would something like this be done practically anyway. I think the SG3526 does allow current mode control. The turns ratio is 1:7. When I removed the voltage feedback at idle I noticed the voltage shoot up to +/- 92V at 12.6V input. The battery is fully charged. The battery voltage only falls to 10.5 (measured at the CT of trafo) when I think the trafo saturates. Without saturation it stays at 11.2V. (65AH battery).
How would you use a scope to measure current with ordinary probes ?
Eva I was referring to a Full bridge not the half bridge. I know that you will need twice the amount of switches but since IRFZ44's cost here around 70 US cents each it does not seem that unfeasible. Furthermore the way I look at it is that you only need 12 x 1.61mm wires for the Pri and not 24. This should greatly simply the trafo winding process. What do you think?
Can anyone please tell me how much inductance is necessary for the input and output chokes ? I chose values that I saw commonly used. Is there formulae for this purpose so that I don't have to do thumb sucking ?
Can I please ask the gurus out there to take a look at the attached spreadsheet and give me any feedback if possible.
I use it whenever I wind trafo's.
PS. Do you think it will be more feasible to wind a 305.9mm Ac (I have this too) or will the 152.4 Ac give me the required 1000W ?
thanks everyone for your input!
At higher voltages full-bridge is preferred over push-pull, but at 12V push-pull is most cost-efficient solution. To get same rdson-losses you need quadruple amount of similar mosfets for full-bridge compared to push-pull. Actually it would be better to go for 30v mosfets then as they need to withstand only half voltage compared to push-pull.
"How much of copper do you think is practical ?"
I'd suggest 5-6 1.6mm conductors per primary turn as a starting point. One conductor per FET would give a nice, simple layout if you were using extended primary leads. This would likely be enough for virtually any audio amplifier. If you want it to be able to run at full power sine wave output continuously at high ambient temperatures, you would need more.
I'd suggest 5-6 1.6mm conductors per primary turn as a starting point. One conductor per FET would give a nice, simple layout if you were using extended primary leads. This would likely be enough for virtually any audio amplifier. If you want it to be able to run at full power sine wave output continuously at high ambient temperatures, you would need more.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.