Transformer Selection

[Zaaphod]

I am looking for some advice on transformer selection for my first DIY amplifier. I am not trying to blow the windows out of my house or anything, I want to start off small and go from there. I would like it to be fairly effecient, so I don't want to just throw as much power at it as possible.

I am using an LM1875 amp x 2 for stereo. I will be driving 8Ω speakers.. something like 6" coaxial speakers. According to the datasheet, power supply can range from +/- 8V to +/- 30V thats quite a range. Maximum power output is 30W... however I don't need it maxed out all the time.

I did some rough calculations please correct me if my formulae are inacurate.. doing them from memory, and it's been more than a few years

30W into an 8Ω load would draw I=sqrt(P/R) = sqrt(30/8) = 1.94A
And Voltage required for this would be E=P/I = 30/1.94 = 15.46V

Since the voltage is RMS and my output wave would need to get to peak values, I need to convert to peak, or 15.46*1.414 = 21.86v

and since my amplifier, which seems to me to be an opamp with power transsitor on the outputs, I suspect it starts clipping before it gets to the rails, by probably at least 1-2V so I'll give it a few volts to work with and say my minimum voltage requirement is 24V would there be any quality difference going to 30V supplies over 24V, or would the extra voltage only require the IC to dissapate more heat? My thinking is that if the speakers are using 15.46V @ 2A, then the IC would need to absorb the remaing voltage at the same current, so at 24V it would absorb 8.54V @ 2A or 17.08W and at 30V it would have to absorb 14.54V @ 2A or 29.08W, so it would be getting quite a bit hotter. But I am still thinking there was something about transistors being not quite linear across their enire operating range, so you want the power supply to be a bit higher than actually needed so you can stay in a more linear range which would reduce distortion... so the question to those of you who have more experience that I do with this... does a 30V power supply sound better than a 24V?


Now to the transformer selection question.. Whatever powersupply voltage I decide on, how do I pick out the transformer? The caps will charge to peak, but they will drop closer to the RMS value when loaded.. and I don't want to burn out my chip which has an absolute maximum voltage of 30V, so if I use the maximum voltage and work backwards, I would need a transformer with 21.22v outputs, which will charge the caps to 30V, or do I put some kind of load on my power supply output so it won't quite charge the caps all the way to peak? I notice a lot of powersupplies have a resistor in series with the final output cap.. this seems like it would be more of noise supression than a load though. there are usually bleeder resistors across the caps as well, but this isn't really enough resistance to drop the voltage any significant amount.

also what size VA transformer do I want for this? here I am getting confused.. My speaker will use say 2A at full load, however, I'm using a split supply, and each power supply output will only be drivng it's transistor 50% of the time, since this is a class B amp... so does that mean I only need an average on 1A going into each rail... and the caps can charge while the opposie rail is driving the load? or does each rail need to deliver the full 2A because the capacitors will not be enough to supply the current at the moment it is needed, even if it is only 50% of the time? Also my 2A estimate was an RMS value calculation.. do I need to convert this to peak as well, or do the capacitors take care of that when needed?

How do I calculate VA? is it (VoltsxAmps)/Power factor? what is the power factor of regular E I transformers? and how about torroid core transfomers? I notice a LOT of people on here use torroid core transformers.. what is the reason for this? do the maintian a more stable voltage than E I transformers? Are they more effectient at high power levels? or do they just look cool?

So thats as far as I got... so this is what I figure I need:
dual secondaries
somewhere between 20 and 30v
be able to deliver between 1 and 2 amps,
and that would be some size VA transformer, depending on what the formula really is and if power factor is part of it, what the power factor acutally is..
and this will be some type of transformer either EI or torroid.

Thanks for any advice anyone can offer on this subject

[mnemneth]

First up, I think you're being a bit ambitious using the LM1875 for a 30W amp.

If you have a closer look at the datasheet, you'll note that for 30W out, you absolutely DO need a +-30V supply.

Further to that, the datasheet also states the MAX current output at Vs - 10V is limited to 4A and with an 8R load that might very easily drop to 6R or less, you will, IMHO certainly be in the limiting range.

I would suggest a different amp chip for 30W or reduce your expectations for your circuit.

Other than that, the very best of luck with your project.

Sandy.

[indianajo]

EI transformers are expensive now. The old triad-utrad plant has been plowed under; they are still for sale at newark.com from c**** but look at the prices. I bought one for my battery charger that was struck by lightning, only 6A and nearly $80. However, I don't dare buy one of those modern IC tricky bits of silicon battery chargers, it would blow up if there was a storm in the next county. Now look at antekinc.com for toroids. Somebody here recommended them. I've been finding surplus and not bought their stuff yet, but the numbers look okay.
I have a 1966 design amp where the Peak rail voltage (90V) exceeds the rating on the power transistors (60V) , and they pulled the rail down with a bypass regulator. Vendors don't regulate PS rails much anymore. I don't know why. Buying transistors with Vceo voltage up to the rail is no longer a problem- in discrete transistors the Vceo is way higher than necessary to allow the Safe operating area to be useful. IC's amps is another area I know even less about, except many are using LM3886.

[AndrewT]

Read up on how to determine the DC voltage one can obtain from a particular transformer.

Start with the transformer specification and the mains electricity Supplier's specification.

eg.
230:20+20Vac transformer with 7% regulation when powered from a 230Vac mains supply will output 20+7% & 20+7% from the dual secondary.
and move forward from that starting point.

I estimate +-32.7Vdc from that 20Vac transformer when very lightly loaded.
Will your 1875 chipamp blow up, or shut down, or amplify within specification?

[Zaaphod]

Originally Posted by mnemneth

First up, I think you're being a bit ambitious using the LM1875 for a 30W amp.

I am not trying to build a 30W amp, I'm trying to build a cheap little amp that sounds reasonably well at fairly low levels. I just want the sound quality to be good when it is putting out say 5W of power. If I was being power hungry, I would use the LM3886 and throw a montster power supply on it, but thats not my goal. The reason I want to figure this out is because I would like to eventually have LOTS of little amps running little speakers all over the place.. in certian areas I will have big amps with more volume, but in most places I want little amps... Since I am planning on making a bunch of these, I want to see what the minimum I can get by with that will still have good quality at low volume levels. I want to make these inexpensivly as possible, however I don't want my prototype to be lacking in it's potential sound quality because I went too small on the transformer. I figure I will build a prototype to the maximum potential of an LM1875 then see what level I will ACTUALLY run it at, then reduce whatever I can to maintain good sound quality at whatever that level is.

as others have mentioned, transformers can be expensive, so that is why I am trying to figure out whats the minimum transformer I can get that will still run the LM1875 at its full potential

[Spiny]

I'd go for an 18-0-18 or 0-18 0-18 transformer, this will give @ +-29V max. with high mains and low loading. Running the thing at 1-5watts will not knock the PSU voltage back much.
VA - in watts you will need 25/30 watts so I'd use around 40va. Others more knowledgeable than me can do the calculation. Use a toroid type.

If you wish to try at a higher voltage or bring the voltage up because the local mains is low then adding a few turns onto a toroid is easy and will add a few volts for testing.

The spec sheet on this chip suggests distortion kicks in anyway at 25watts (1%THD) so I'd keep to a lower voltage than the max, will help with heat dissipation as well.

[Zaaphod]

Originally Posted by AndrewT

I estimate +-32.7Vdc from that 20Vac transformer when very lightly loaded.

Originally Posted by Spiny

I'd go for an 18-0-18 or 0-18 0-18 transformer, this will give @ +-29V max.

Thank you both for your help, I am getting a good idea on what to do here.

Can you please explain how you guys are getting these figures?

for a 20v transformer, I get a maximum unloaded dc voltage of 28.284VDC on the caps when they charge to peak. I use the formula Vpeak=Vrms*1.414
actually I will only get 26.88volts out of my power supply, because diodes have a 0.7volt drop each, and in a full wave rectifier two diodes are in series with the capacitors at all times.

for 18V transformer I get 18*1.414=25.45V Peak Voltage - 1.4V for the 2 diodes = 24.05V

I will explain my formulae...
transformers list the RMS voltages
Vrms=Vpeak/SQRT(2) and square root of 2 is 1.414.. so if you know the RMS value, transpose the formula to

Vpeak=Vrms*1.414

this is the highest value you could charge a cap to, if you had ideal diodes, but real diodes drop 0.7V each, so you must subtract 0.7V for each diode in the circuit.. since a full wave bridge uses 2 diodes at any given time, and alternate between 2 sets of 2 diodes as the cycle changes, we must subtract 0.7V *2 or 1.4V for the diode voltage drop, thus, the maximum charge on a capacitor for a full wave bridge is

VDC = (Vrms*1.414)-1.4

For a full wave center tap design, I would only need to subtract 0.7V because only one diode is in use at a time, but since I want a split supply with one transformer, a full wave bridge on each of 2 separate secondary windings would be best.


looking at the spec sheet on a few transformers, it looks like the voltage listed is the unloaded voltage, and it will only drop from there... I'm not sure if that is how they are all listed however.

I am looking at
Antek - AN-0520
the data sheet lists
20.2V at no load
19.7V at 19W output
18.8V at 43W output

This seems to be 6.9% regulation which seems pretty good to me for a $13 transformer.

they have an 18V version of it as well, I might still go with 18v.. I am just curious about the math involved here. I just can't see how you guys are getting such high voltages, but since you both came up with numbers much higher than I did, I want to see what is going on.

[AndrewT]

I asked you to read up on transformer to DC PSUs.

You didn't !

Allow for transformer regulation.
Allow for mains electricity supplier voltage tolerance.
Both of these were specifically mentioned.

Start with the transformer specification and the mains electricity Supplier's specification.

if this sentence did not make sense to you then you could ask for more detailed information.
Why ask again when you choose to ignore?

[Zaaphod]

i read this
Electronic Circuits Design For Beginners - Chapter 1 - a knol by Max Iskram

[mnemneth]

Originally Posted by Zaaphod

For a full wave center tap design, I would only need to subtract 0.7V because only one diode is in use at a time, but since I want a split supply with one transformer, a full wave bridge on each of 2 separate secondary windings would be best.

20.2V at no load
19.7V at 19W output
18.8V at 43W output

This seems to be 6.9% regulation which seems pretty good to me for a $13 transformer.

they have an 18V version of it as well, I might still go with 18v.. I am just curious about the math involved here. I just can't see how you guys are getting such high voltages, but since you both came up with numbers much higher than I did, I want to see what is going on.

Actually, your math is correct. I too don't quite get where the others are getting their numbers from. No transformer manufacturer I've ever known quotes regulation as a "positive" value. They ALWAYS quote best possible output levels, i.e. if they say it's a 20 volt output, then on a scope with NO load, I'd expect to see 56.5V Pk/Pk, NEVER higher.

For a 30V Peak supply with 7% regulation, you will need 32.1V peak.

However, when you take diode drop into consideration, this rises to 32.8V, say 33V to be safe.

Therefore your "RMS" transformer rating will need to be 33/1.414=23.3V.

I'd go for 24V, as that's a common value and should be reasonably cheap to buy.

As for your "split" transformer, I presume you want to power the left and right channels separately from one secondary each?

In which case, you'll need to subtract 2 diode Vbe's from your output voltages.

Therefore, I would still recommend a 24V (RMS) secondary output.

Sandy.