Point to point prototypes are ok as long as you remember that they will work well only at low output powers and they will provide higher ringing and more EMI than a properly designed PCB
Anyway, try allways to keep the main AC current loop small, ie: input capacitor, MOSFETS, RC dampers and transformer primaryes. Try also to keep the output AC current loop small, ie: secondaries, RC dampers, rectifiers and filter capacitors [for unregulated supplies this is less critical due to leakage inductance seriously limiting the maximum dI/dt on that loop]
Try to place the SG3525 near the MOSFETS and ground it near its drains. The lower the gate resistors the lower the maximum area allowed for gate drive loops without Vgs ringing and posible parasitistic turn on
Anyway, try allways to keep the main AC current loop small, ie: input capacitor, MOSFETS, RC dampers and transformer primaryes. Try also to keep the output AC current loop small, ie: secondaries, RC dampers, rectifiers and filter capacitors [for unregulated supplies this is less critical due to leakage inductance seriously limiting the maximum dI/dt on that loop]
Try to place the SG3525 near the MOSFETS and ground it near its drains. The lower the gate resistors the lower the maximum area allowed for gate drive loops without Vgs ringing and posible parasitistic turn on
hello everybody
there are very few smps circuits for amps and most of them are based on the sg3525 , if u will follow this thread from the beggining u will see that i made the smps with a uc2825 chip , this one got 2A peak current so u dont have to add no buffer
i made the smps without using an osciloscope ! so i got a few problems like voltage spikes .i'm keeping money for the osciloscope and after i'll buy it i'll publish the circuit .
anyways what i'm trying to say is forget the sg3525 and use the uc2825
there are very few smps circuits for amps and most of them are based on the sg3525 , if u will follow this thread from the beggining u will see that i made the smps with a uc2825 chip , this one got 2A peak current so u dont have to add no buffer
i made the smps without using an osciloscope ! so i got a few problems like voltage spikes .i'm keeping money for the osciloscope and after i'll buy it i'll publish the circuit .
anyways what i'm trying to say is forget the sg3525 and use the uc2825
B.I.G said:
and you are probably generating loads of RFI -- this isn't a trivial problem. a poorly designed SMPS is like running a spark-gap transmitter.
U.S rules are strict, and the E.U. more restrictive still -- and for very good reasons -- a poorly designed SMPS will raise the noise level for RF communications for police, fire, ambulance services, it can disrupt the operation medical equipment, etc.
and for us audio enthusiasts unwanted EMI causes fluky currents on input leads, power supplies etc.
simple way of testing your SMPS -- use an inexpensive battery operated a.m. transistor radio -- you don't need a $10k Agilent spectrum analyzer to see if there's a problem.
Here are some other methods which I use. One way I also find convenient is tuning in the local broadcast TV channel in the lower half of the VHF band. Then EMI noise shows up as static in the picture or even in the audio if there is an even greater problem. The different patterns are interesting to examine, and then subtle changes in them may be observed after trying various circuit modifications. I know a circuit has a severe problem if my DMM begins to squeal over-voltage warnings when either connected or not to the circuit. The medium sensitivity indicator is a cheap FM stereo from the older days which I always have operating. I don't like to hear any interference on it.
Be sure to include a low ESR cap from the drain of the upper to the source of the lower MOSFET when the circuit uses totem pole (half bridge or bridge) transformer drive. My general feeling is that if the leads of the cap do not reach these locations by themselves, then the high frequency, high power loop is too expansive, for p2p circuits anyway, and old oversize-bodied caps do not count.
Some mylar caps are better than others. Using twisted wire runs may help also. CMMR filtering may be necessary at lead inputs and outputs of the metal enclosure. Pre-made line entry filters may come in handy. I have heard of using personal computer PS cases for housing projects.
subwo1 said:
you can build the SG3524 and SG3525 as "sub-circuits" -- all the information is in the data-sheets -- it's basically a voltage reference, an two error amplifiers, one comparator, a multivibrator, two flip-flops ... what will mystify you even with a spice model is the compensation for the comparator/error amplifier --
A grounded metal case provides decent attenuation to electric fields [capacitive propagation of quick voltage changes] but attenuation to magnetic fields [inductive propagation of quick current changes] may pass through the case with 20dB or less attenuation
Also, the case does nothing against EMI radiated through the input and output wires acting as antennas
Also, the case does nothing against EMI radiated through the input and output wires acting as antennas
What makes a good smps?
Hi,
What in your experience makes a good SMPS for audio use.
Looking at an off line converter -
Active PFC with more than enough 'C' to cover low line worst case?
Good LF gain?
tightest regualtion at the output?
loop response that covers the audio bandwidth?
Transient response?
Low output noise?
etc.
Just looking form some parameters and what benefots they offer over a traditional linear PSU for use in a domestic amplfiier
Cheers
Sheriff
Sheriff
Hi,
What in your experience makes a good SMPS for audio use.
Looking at an off line converter -
Active PFC with more than enough 'C' to cover low line worst case?
Good LF gain?
tightest regualtion at the output?
loop response that covers the audio bandwidth?
Transient response?
Low output noise?
etc.
Just looking form some parameters and what benefots they offer over a traditional linear PSU for use in a domestic amplfiier
Cheers
Sheriff
Sheriff
I have simulated my design of a SMPS because it is a very simple circuit theoretically. I know it is not perfect, but I have not found an approach I like more, even in real life. That circuit definitely shows the benefits of phase compensation in the feedback loop. I think that concept is what you mean.jackinnj said:
smps and class d
Hi Subwo1,
How about the pros / cons compared to a linear PSU powering a class D amp for example.
I know you have much experience on this topic also.
I guess that with a bit of messing around you could sync an active PFC stage to the downconverter to a multiple of the Class D amps switching frequency in an atempt to avoid beat frequencies.
The trouble is developing a good amplifier with a fixed carrier frequency....., and then filtering the sync'd noise...
Cheers
Sheriff
Hi Subwo1,
How about the pros / cons compared to a linear PSU powering a class D amp for example.
I know you have much experience on this topic also.
I guess that with a bit of messing around you could sync an active PFC stage to the downconverter to a multiple of the Class D amps switching frequency in an atempt to avoid beat frequencies.
The trouble is developing a good amplifier with a fixed carrier frequency....., and then filtering the sync'd noise...
Cheers
Sheriff
Hi Sheriff,
Reality turns out to be not much more than a bunch of trade-offs. I try not speak on the issue of the best approach for any one individual, unless they already have a set course I can give advice I am comfortable with.
I did tinker with an unregulated SMPS which also was the source of the constant switching frequency of an amp. But other projects needed attention and I did not return to experimenting with the circuit.
Reality turns out to be not much more than a bunch of trade-offs. I try not speak on the issue of the best approach for any one individual, unless they already have a set course I can give advice I am comfortable with.
I did tinker with an unregulated SMPS which also was the source of the constant switching frequency of an amp. But other projects needed attention and I did not return to experimenting with the circuit.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.