General Purpose 20MHz Function Generator Amp

Complementary emitter followers on a heat sink, and a couple of bench supplies?

Two diode drops between gates will do for forward bias.

No need to insulate the paks from the heatsink, as the source terminal is the tab, so the heatsink is at output voltage.

A+/- 15 volt supply limited to say 1A should do fine.

... Phil

Does it just need to sink current? Can you handle a bit of attenuation and voltage shift? If so, why not just a PNP or P-channel FET?

Diodes. I hadn't seen that used before. I will try it.

Barry Latham

Just saw this in a mag....

You might want to take a look at the power op amps from Apex Microtechnology. They have several models which would easily meet your output current requirements and appear to be fast enough. Some of them are described as being suitable for driving motors or piezo drivers; they have built-in current limiting, inductive-kickback clamping diodes, and thermal shutoff.

The PA119 would be one possibility... 100 MHz gain-bandwidth product,

8-pin TO-3 case. Use a good heatsink :-)

Not cheap (about $300 in onesies) but it would be fairly straightforward solution. Some of their smaller, lower-power parts might suffice for your needs.

$1875 for the cheaper model that does 16Vpp into 50 ohms. Several time I wanted more out of my sig gen, but not at that price. Mikek

OK, that is expensive. Maybe I should do one as a product.

We sometimes need to make 26VRMS for synchro simulation, and sometimes 120 RMS, from a low-voltage ARB, but that's a different product.

Interesting! Do you have a guess as to what your cost would be if you did?

not 500mA but in the ball park.

I almost made that suggestion when I found the price. Mikek

A similar device was previously discussed in this newsgroup. The best suggestion was a pair of LT1210 devices in the form of a bridge amplifier for 9 watts into 50 ohms to about 14 Mhz. Note that it's NOT DC coupled. Search Google Groups in this newsgroup for LT1210 .

Data sheet:

App Note. See Fig 70:

My tweaks:

It's only 0.625 of $3000!

This looks like it could be done with available opamps, maybe a few in parallel at the output. THS6022 or some such. Or something discrete for slower but a bit more voltage swing.

Den tirsdag den 1. oktober 2013 02.36.11 UTC+2 skrev John Larkin:

haven't looked too closely at it but maybe:

-Lasse

That circuti looks worth trying.

Can you possibly recommend a "workable" substitute for the CoilCraft transformer?

Something more commonly availalble or self-made on a generic core.

Thanks,

Barry Latham

"John Larkin" "Phil Allison"

Two FETs as I suggested will fit his spec for under $20 - plus a PSU.

.... Phil

The most critical part of any design is the specification. And yours is beyond vague.

It's not unusual for customers to think vaguely about what they might need, then hedge their bets by asking for 10x that much. The difference between 10 and 20V might make a BIG difference in the cost of the devices. Ditto for 1 vs 10 MHz. And what does 10MHz. mean? Are you content with a 3dB bandwidth of 10MHz.? Or do you want competent square waves at 10MHz. rate? Transient response requirements? Linearity of the triangle? Flatness of a frequency sweep? Flatness of the top of a square wave? DC coupled? Voltage or current offset?

And how are you gonna get the signal to the load? Mismatched cables are not your friend.

It's been over 30 years since I designed a production function generator, but physics hasn't changed much since then. Meeting any few requirements is easy. Meeting all of them simultaneously in the same box is a bitch. "General purpose" only has meaning to the person who made the statement.

You need to be very clear about what you want.

Having said all that, I second the recommendation that you build complementary emitter followers. But there's art in even the simple approach. Thermal time constants can mess up the flatness of the tops of your square waves at low frequencies. Or induce variations in offset that are waveform dependent. The output impedance of an emitter follower has an inductive component that you have to compensate for. You're probably gonna want some current limit. Overcurrent shutdown is a lot simpler than a smooth transition to current limit in active mode. Carefully consider the load fault current when selecting devices that may have safe-operating-area issues. You can also get blind-sided by SOA issues when driving inductive loads.

Be aware that this isn't a current amplifier. It's a voltage amplifier with a gain less than one and lower output impedance.

If your objective is to supply current, there may be better ways.

"VPH5-0155-R" That seems to be a Coiltronics part number.

You could possibly substitute Coilcraft HPH5-0155L

Hit their website