Question: SMPS switching frequency with 41Hz Amp4

DVBjunky

Member
2006-12-30 2:01 pm
Hi,

I am planning a small Tripath based Amp with the 41Hz Amp4 Kit but I am stuck with some questions about the SMPS?

1) I found some cheap MeanWell SMPS with 100W 24V 4,5 Ampere but with a switching frequency of 25 Khz. Is that too low?

Do I have to buy an SMPS with 50Khz, 63Khz or even 100Khz switching frequency?

2) Is 4,5 Ampere enough? 24V rail with 4 Ohm speaker? I have 89db effiency speaker.

3) Are they any cheap SMPS outhere? I Mean cheap and not Coldamp! :)

Thanks and regards

Alan
 
@DVBjunky:

1) Any switching frequency above the audible frequency should be acceptable - harmonics are multiples of the switching frequency. :)

2) The answer is a little tricky: For 4 ohm loads Tripath recommends to parallel the outputs caused by the current limits of each channel but I guess the AMP4 isn't designed by Jan to have this option.
At 23.5V and a 6 ohm load you get just 23W (THD+N = 0.01%) at each channel, so a 108W SMPS (24V/4.5A) would do it easily. Assuming an efficiency of >80% you just need (23W + 23W) / 0.8 = 57.5W. :)

3) My source for cheap stuff is Pollin.de. You might give it a try. They have once in a while good bargains about SMPS.

Your last question) 20V is the lower end for the Tripath TK2050 chipset where it feels comfortable with the power supply voltage - however with further descreased output power. Therefore a 90W SMPS (20V/3.5A) will do a good job though. :)

Besides, If you connect a SMPS to the AMP4 make sure that the SMPS is stable when running on extremely large capacitiv loads (2x 10000µF(!!!) bulk capacitors). In case the SMPS acts a little "strange" (output voltage start to oscillate or, more likely, the SMPS just turns off when rapidly drawing current, because of current limit protection - if there is any protection scheme at all) you might have to remove the two 10000µF bulk capacitors.
 

DVBjunky

Member
2006-12-30 2:01 pm
Hi,

thanks for the advice.

I missed the last Cheap Meanwell SMPS offers on Pollin last year. Damn...they were really cheap....a bargain.

Do you really think I have to remove the 2x 10000uF Caps or should I lower the value of the caps.......I am about to start the soldering, so it would take 2-3 weeks till I get there :)

Tanks and regards

Alan
 
@pafi:

In BTL (Bridge Tied Load) configuration you don't have any power supply pumping effect :whazzat:, except if the layout and wiring is badly designed - what I doubt if it concerns Jans skills.
That's by the way a reason why a design in theory and in simulations runs almost perfect but in a real and more practical design almost everything rises and falls with a good/bad PCB design. :D

Power supply pumping most likely happens at single ended amplifiers and at low output frequencies only - then the "revers" current must be "buffered" by the bulk(y) caps. :eek:

In a BTL design the "revers" current will be used by the other side of the bridge. :D Therefore it is necessary to have a short and low impedance track between the power supply tracks of the (+) and (-) output stage of each channel.


@DVBjunky:

If you're in doubt try to get a full datasheet of the SMPS your gonna use. In many (if not most) cases the manufacturer states the maximum capacitive load allowed for a stable and reliable operation of the SMPS.
In all my decades of experience in repairing/using/abusing SMPS I've seen just one SMPS which starts to oscillate as soon as the reactive load (capacity/inductivity) exceeds a certain level. Most likely current SMPS designs will limit the value of the max. output current. If the SMPS runs to long in that limit it expects a short at the output and just shuts down. How a certain SMPS will really react depends on the topology, the involved PWM controllers, and many other factors I won't list here.
 
Corax!

You are almost right, the strongest kind of supply pumping effect occures at SE amps, but not exclusively! Speaker load is not ohmic, but reactant (especially near resonance of woofer). Direction of energy transfer reverses for some part of period, and if it exceeds losses, then supply will start to increase, because most of SMPS-es are unable to sink current.
 
Damn, I have to apologize. That happens if you don't read your post a second time from the beginning on before you hit "Submit Reply".
I started that sentence with another formulation, decides then to edit it, deleted a few words, added a few words, .......... :headshot:

Of course I won't express that you don't have power supply pumping at all when using BTL configuration. The effect is always there but much less than with single-ended designs.

To lower the effect (hence the voltage rise) the only chance is to increase the bulk capacitor value(s).
The issue with SMPS is that they're not always stable with high reactive loads. That's the main point I wanna mention.

I've seen some professional lab power supplies (i.e. from Agilent - formerly known as Hewlet Packard (HP)) with such a "bleeding" circuit that prevents the rise of the output voltage when attached to reactive loads and more or less fast varying output voltages. In most cases the power dissipation of the "bleeding" circuits is much weaker than the "driving" part of the output.

Anyhow thanks for your correction Pafi. :up: I really appreciate it and we won't pass on wrong informations to other DIYers. :xfingers:
 
Hi Pafi!

How can the supply voltage increase if the speakers are passive components? Assume realistic output capacitor values.
I think about the controller of the supply will decrease the power the output effectively if it has a steep characteristic, and more if has pole(s) at zero.
Do you think a quarter period of output current can overcharge the supply caps?
 
Sorry, I don't have time to explain this step-by-step. The key is energy transfer (since PWM amps have almost 100 % efficiency), and reactancy.

Controller usually don't do anything, because they are designed for positive power output. (There are exceptions.)

Do you think a quarter period of output current can overcharge the supply caps?

Not surely over, just charge.
 
Pafi said:
Controller usually don't do anything, because they are designed for positive power output.

How can the controller of a simple Class-D output be designed for negative power output? Otherwise said, is the design method of a controller for a VCM PSU in CCM different from a controller design for a VFB Class-D output?

Pafi said:
Controller usually don't do anything

What doesn't do a controller of a VCM PSU with the PWM signal, at negative power output, from that a controller of a VFB Class-D output does?
 
Gyula!

Bocs, nincs idõm angolul szavakat keresgetni! Te vetetted föl a vezérlõt, nem én! Én csak belementem a hibás alanyodba az egyszerûség kedvéért. Nem volt idõm kinyomozni, hogy vajon mire is gondolhatsz, és hogy jön ez ide, aztán kijavítani. Most ne köss belém azért, mert az általad felvetett dolognak semmi köze a témához!

Most meg megint olyan dologról kezdtél írni, aminek szintén semmi köze a témához, hiszen senki sem beszélt a táp eredeti kondenzátoránál kisebb beépítésérõl!
 
Pafi said:
Gyula!

Bocs, nincs idõm angolul szavakat keresgetni! Te vetetted föl a vezérlõt, nem én! Én csak belementem a hibás alanyodba az egyszerûség kedvéért. Nem volt idõm kinyomozni, hogy vajon mire is gondolhatsz, és hogy jön ez ide, aztán kijavítani. Most ne köss belém azért, mert az általad felvetett dolognak semmi köze a témához!

Most meg megint olyan dologról kezdtél írni, aminek szintén semmi köze a témához, hiszen senki sem beszélt a táp eredeti kondenzátoránál kisebb beépítésérõl!

What? I don't understand! :) Bocs, most olvastam el a topicot az elejérõl. Én arra gondoltam hogy ki akarja venni az elektrolit kondenzátort a kimenetrõl, fóliának meg illik még lenni benne a kis ESL-hez, a töltõ áram felfutásnál.
Az áram visszatápláláson kívül a stabilitásra is figyelni kell. Most látom hogy az Amp4-ben 10000 uF-os kondik vannak. Akkor viszont elõfordulhat hogy mégis hozzá kell nyúlnia a szabályozóhoz, mert a VFB zérussal állítja be a fázistartalékot a másodfokú szûrõ miatt, amit így kisebb frekvenciára kell rakni. Már ha nem méretezték akkora tartalékra. Szerintem érdemes lenne megvizsgálnia a tápegységben és az erõsítõben lévõ kondenzátorok értékét. Ha a kettõ összemérhetõ, akkor még szorosabb gyári méretezésnél is van esély a stabil mûködésre.


Pafi said:


Yes, rectifier is not controller, but S. rect. needs control. Actually I didn't want to state anything about controller originally


Ok! My controller is the regulator from now. ;)


Pafi said:


I didn't want to discuss about at all!



Corax, DVBjunky, I think if the supply hasn't got a datasheet which contains the maximal capacitive load, you should to see into the PSU's output capacitor value.
Because if the supply utilizes a VMC regulator, the gain and phase margins are set with a zero and a faster pole, because of the second order output filter. If the capacitor value is enlarged, the cutoff will happen at a lower frequency, so the zero (and maybe its neutralizing pole) also should be moved to lower frequency, if the factory design was tight. I think it would be worthy to compare the PSU's output capacitor and the Amp4's supply capacitor. I think if the Amp4's capacitor value is not, or not so larger than the PSU's, there is a good chance for stability even with tight factory design.

If the PSU utilizes a CMC regulator, enlarging the output capacitor value is not hit the stability. The supply should be stable.
 
mert a VFB zérussal állítja be a fázistartalékot a másodfokú szûrõ miatt

Ez nem jellemzõ. Az elkó nagy frekin ellenállás, tipikusan ide méretezik a vágási frekit, így a szûrõ csak 90 fokot tol, sima P szabályzó (a precizitás miatt PI) elég. De most már nem offolok tovább.

Ok! My controller is the regulator from now.

I have no idea what you are talking about.