(audio) "PA amp"?

[ge]

We want to test a design at AC line frequencies from (say) 50-70 hz.
Cost is an issue. The unit draws about 20W. Line is 120V.

One suggestion we've gotten is to feed a sine-wave generator into a
"PA amp", and just step that output up with a variable transformer.
Supposedly, these "PA amps" can swing 70v (rms?).

I'm having trouble finding specs on these devices. It's not clear
from what I find with Google even whether any particular amp has this
70v output.

In general, I'd like a pointer to some sort of technical overview of
the land of audio power amps, which I pretty much know nothing about.
In particular, I'd like to know if they have strict load impedance
requirements, or any other gotchas.

And, of course, if you think this sounds like a not-so-good idea, it
would be OK to say that.

TIA,
George

 

[John Larkin]

We want to test a design at AC line frequencies from (say) 50-70 hz.
Cost is an issue. The unit draws about 20W. Line is 120V.

One suggestion we've gotten is to feed a sine-wave generator into a
"PA amp", and just step that output up with a variable transformer.
Supposedly, these "PA amps" can swing 70v (rms?).

I'm having trouble finding specs on these devices. It's not clear
from what I find with Google even whether any particular amp has this
70v output.

In general, I'd like a pointer to some sort of technical overview of
the land of audio power amps, which I pretty much know nothing about.
In particular, I'd like to know if they have strict load impedance
requirements, or any other gotchas.

And, of course, if you think this sounds like a not-so-good idea, it
would be OK to say that.

TIA,
George

Any audio amp would work. I'd sugggest a 50-watt or so amp, just to
have good margin. A garage-sale stereo receiver should do fine. I use
a Peavey 400 watt/channel stereo PA amp to test electrical metering
products... one channel steps up to 120-480 volts, and one steps down
to 0-200 amps to drive current transformers. With a stereo amp, you
can bridge the channels and get (almost) twice as much power if you
need it.

The idea of using a Variac backwards is a good one. That lets you
optimize the impedance match.

John

 

[Walter Harley]

Any audio amp would work. I'd sugggest a 50-watt or so amp, just to
have good margin. A garage-sale stereo receiver should do fine. I use
a Peavey 400 watt/channel stereo PA amp to test electrical metering
products... one channel steps up to 120-480 volts, and one steps down
to 0-200 amps to drive current transformers. With a stereo amp, you
can bridge the channels and get (almost) twice as much power if you
need it.

The idea of using a Variac backwards is a good one. That lets you
optimize the impedance match.

To figure the transformer size, first work out the voltage (rather than
wattage) that the amp can produce. Stereo amps are often rated into 8 ohms;
so, if an amp can generate 50W into 8 ohms, it's putting out V = sqrt(P*R) =
20V. To get 120V out the transformer, you want to use a transformer with a
120V primary and an 18V secondary, in reverse. (Power transformers are not
100% efficient; that's why I said 18V rather than 20V.)

Next, remember that most inexpensive amps are rated for "music power" rather
than continuous sine wave output. If your device draws 20W, that means you
need to feed the transformer about 25W (again, not 100% efficiency), which
means you need an amp rated for at least 100W so that it won't overheat.
Some amps derate by as much as 10x for continuous low-frequency sinewave
power. Larkin's Peavey is an excellent choice.

If using a Variac, be aware that the amp's output is not isolated from the
transformer output. If your device itself has a transformer input, you're
probably fine. But in principle, you could get a surprise, for instance if
you used the amp in bridged mode and then you tried to measure the voltage
across the device input with a non-isolated oscilloscope. Normal power
precautions apply.

 

[Mike Page]

Any audio amp would work. I'd sugggest a 50-watt or so amp, just to
have good margin. A garage-sale stereo receiver should do fine. I use
a Peavey 400 watt/channel stereo PA amp to test electrical metering
products... one channel steps up to 120-480 volts, and one steps down
to 0-200 amps to drive current transformers. With a stereo amp, you
can bridge the channels and get (almost) twice as much power if you
need it.

Yes, get a PA amp. Don't expect a HiFi amp to be as robust, and steer
well clear of automotive amps, they're specced in silly numbers. Watch
out for earth loops.

You can string a AGC round it to compensate for droop with the smaller
transformers.

Regards,
Mike.

 

[John Larkin]

Yes, get a PA amp. Don't expect a HiFi amp to be as robust, and steer
well clear of automotive amps, they're specced in silly numbers. Watch
out for earth loops.

You can string a AGC round it to compensate for droop with the smaller
transformers.

Regards,
Mike.

Oh, one other trick: a capacitor across the 120 side can reduce amp
loading and make the drive nicer for the DUT.

John

 

[NCSRadio]

With a stereo amp, you

can bridge the channels and get (almost) twice as much power if you
need it.

John

John,
Don't you get almost 4x the power with a bridge because P=E^2/R ?
Carey Fisher

 

[John Larkin]

John,
Don't you get almost 4x the power with a bridge because P=E^2/R ?
Carey Fisher

Depends on the power supply and heatsinking. If either of them is the
limit - and one generally is - you can't get 4:1. They're both
expensive, so amp manufacturers are not likely to provide twice as
much as absolutely needed, unless they claim the bridging feature
maybe.

John

 

[Pooh Bear]

We want to test a design at AC line frequencies from (say) 50-70 hz.
Cost is an issue. The unit draws about 20W. Line is 120V.

One suggestion we've gotten is to feed a sine-wave generator into a
"PA amp", and just step that output up with a variable transformer.
Supposedly, these "PA amps" can swing 70v (rms?).

I'm having trouble finding specs on these devices. It's not clear
from what I find with Google even whether any particular amp has this
70v output.

In general, I'd like a pointer to some sort of technical overview of
the land of audio power amps, which I pretty much know nothing about.
In particular, I'd like to know if they have strict load impedance
requirements, or any other gotchas.

And, of course, if you think this sounds like a not-so-good idea, it
would be OK to say that.

I've done this. Don't bother with amps with 70 or 100 Volt line outputs
since you need to step up further anyway.

The load reactance means that you'll need a rather larger wattage
amplifier than the load VA might suggest.

Any typical modern 'PA' amplifier will drive down to a 4 ohm load and
possibly 2 ohms with no upper limit so impedance isn't really an issue.

I assume you know how to calculate the amplifier's output voltage from
the rated power spec ? Then find a suitable transformer and use it in
'reverse' - i.e. connect the amplifer to the secondary and power your
device from the 'primary'.


Graham

 

[Pooh Bear]

Any audio amp would work. I'd sugggest a 50-watt or so amp, just to
have good margin.

My experience would suggest maybe a larger margin might be advisable. The
idea of phase angle correcting with a cap is cute though.


Graham

 

[Tony]

Oh, one other trick: a capacitor across the 120 side can reduce amp
loading and make the drive nicer for the DUT.

John

Which introduces the subject of reactive loading, which linear amps
really hate (and often blow up just to demonstrate that point). At
50Hz with a capacitive or inductive load, the output BJTs can spend a
ms or two going through conditions of high voltage AND high current at
the same time, so it would be a good idea to oversize everything, and
make sure magnetizing currents are well below the nominal max load
current.

Before connecting an amp to a transformer, it's also wise to make sure
the amp's DC offset is tiny (+/-10mV should be fine), otherwise you'll
get potentially high DC current into the transformer's primary.

Tony
Tony (remove the "_" to reply by email)

 

[Guy Macon]

We want to test a design at AC line frequencies from (say) 50-70 hz.
Cost is an issue. The unit draws about 20W. Line is 120V.

One suggestion we've gotten is to feed a sine-wave generator into a
"PA amp", and just step that output up with a variable transformer.

Fixed, not variable. You vary the voltage at the sine-wave generator.

0.2A at 120V is 24W, which tells me that your load is around
600 Ohms at 60Hz. Get a 100:1 step-up transformer and that
0.2A becomes 20.0A, the 120V becomes 1.2V and the load becomes
6.0 Ohms. The power stays at 25W.

A Carvin DCM150 150W Amplifier puts out 50W into 8 Ohms and
75W into 4 Ohms - and that's just one channel. Price: $230
Make that 100:1 transformer a 100W model and you will have
plenty of power for startup surges.

See http://www.carvin.com/products/dcmseries.php

 

[Tony]

Fixed, not variable. You vary the voltage at the sine-wave generator.

0.2A at 120V is 24W, which tells me that your load is around
600 Ohms at 60Hz. Get a 100:1 step-up transformer and that
0.2A becomes 20.0A, the 120V becomes 1.2V and the load becomes
6.0 Ohms.

Typo alert - should be "0.06 Ohms" - not friendly to most power amps.

Tony (remove the "_" to reply by email)

 

[Mike Page]

Before connecting an amp to a transformer, it's also wise to make sure
the amp's DC offset is tiny (+/-10mV should be fine), otherwise you'll
get potentially high DC current into the transformer's primary.

Tony
Tony (remove the "_" to reply by email)

I made the mistake of capacitively coupling the LV winding of a booster
transformer to avoid potential DC saturation currents. I blew several
fuses. I never bothered to analyse why, but replaced the caps with
smallish resistors, which kept the DC current to adequate levels.

Mike.

 

[Pooh Bear]

Typo alert - should be "0.06 Ohms" - not friendly to most power amps.

I think he forgot impedance relates to the square of the transformer turns.

So If you wanted a 6 ohm load - you need a 10:1 ratio.

The output voltage would be 12 V.

I still reckon that the reactive component of the load will require
consideration as will recifier conduction time in the PSU which normally
only conduct for 1/3 or less of a cycle.


Graham

 

[Pooh Bear]

I made the mistake of capacitively coupling the LV winding of a booster
transformer to avoid potential DC saturation currents. I blew several
fuses. I never bothered to analyse why, but replaced the caps with
smallish resistors, which kept the DC current to adequate levels.

Maybe you made a resonant load ?


Graham

 

[Guy Macon]

I think he forgot impedance relates to the square of the transformer turns.

I am not sure why I made such an emementary error, but it wasn't from
forgetting the relationship between impedance ratios and turns ratios.

I blame George Bush.

 

[Nicholas O. Lindan]

We want to test a design at AC line frequencies from (say) 50-70 hz.
Cost is an issue. The unit draws about 20W. Line is 120V.

Instruments are built for this:

http://www.hotektech.com/EPSAmplifier.htm

As an example -- there are much smaller units available, and from
different manufacturers.

A 1 month rental on a small one should be in the $100 range.

But I agree, for 20W a rugged audio amp and a step-up transformer
(ie: PA amp) may be a fine way to go.

I have not found line frequency to be much of a problem, though.
I have found low line testing at 80 volts, cycle drop-out testing
and surge withstand capability are far more important when it
comes to insuring equipment works through routine power line
problems.

* * *

OART:

In 90% of clients that blame all their intermittent problems
on the power company, I have found the true cause to be either,
and often both:

o Ones with really shoddy design, though not in the power supply
o Ones with bad software

The other 10% have a badly designed power supply.

In 27 years of consulting, most of it fixing reliability problems,
I have never found the power company is to blame.

 

[Guy Macon]

In 90% of clients that blame all their intermittent problems
on the power company, I have found the true cause to be either,
and often both:

o Ones with really shoddy design, though not in the power supply
o Ones with bad software

The other 10% have a badly designed power supply.

In 27 years of consulting, most of it fixing reliability problems,
I have never found the power company is to blame.

I had the same experience until I started making things that were
exported to third-world countries.

 

[Pooh Bear]

I am not sure why I made such an emementary error, but it wasn't from
forgetting the relationship between impedance ratios and turns ratios.

I blame George Bush.

I will too in that case ;-)


Graham

 

[Bob Urz]

Tony wrote:




I think he forgot impedance relates to the square of the transformer turns.

So If you wanted a 6 ohm load - you need a 10:1 ratio.

The output voltage would be 12 V.

I still reckon that the reactive component of the load will require
consideration as will recifier conduction time in the PSU which normally
only conduct for 1/3 or less of a cycle.


Graham

Why do i smell smoke at the end of this experiment?

If one were crafty, they could try a Batt op UPS and see if
you could modify its frequency reference generator for variable
operation. These small UPS's are cheap these days.

Bob