Here are a number of things I learned in designing solid state MRI amps with MRF devices:
The bias circuit to a FET is always a high impedance. There are valid RF reasons for making this true. Therefore, it is a simple matter to ground all the FETs except the one whose bias you wish to adjust. In your case, ground 7* of the gates, and the remaining one is what you'll be biasing.
There is a caution here - if this was designed for spectroscopy imaging, then it has sufficient heatsink for CW/SSB. If it was single-frequency MRI, it does not have enough heatsink for much more than 20% duty cycle.
Here is a thermal runaway possibility with a FET, even though most of us have been told there isn't. It's just that it's different. The gate bias voltage required to deliver any given drain current reduces when a FET heats. Therefore, if you set the bias to some nominal value (100mA), then run RF through it and it gets hot, that bias current will go up. I'm assuming they don't have temperature compensation - and my assumption could be very wrong. Therefore:
Check for thermal stability this way: bias the FET to some value and watch the drain current. It will probably continue to rise a bit, then stabilize. That's OK. I probably used 20,000 MRF150s in my work, and depending on design requirements used bias currents per device between 50mA and 1A. Look at it this way: the MRF150 can dissipate about 150 watts on a reasonable heatsink. Multiply your drain voltage by your bias current, and that's the power dissipation at idle. It's probably a good rule of thumb to keep it less than half the power rating, so less than 75 watts. If your drain voltage is 50, then don't bias above 1.5A. And that should be a hard limit - better for FET stability to bias lower.
Another on thermal stability: once you've chosen a bias current and run RF through it, take the RF away but keep the bias going. Now you'll see an elevated drain current, but it should settle back down to quiescent.
Both spectroscopy and single-frequency MRI amps actually are class AB to begin with. However, it is often the case that they required an external signal from the rest of the medical equipment before they would turn the bias "on." Our customer called this "unblank". So look at the circuitry carefully before tweaking the bias pots. In fact, if it were class C, I'd expect it not to have bias pots. You'll probably find some line somewhere that when you ground it, the FETs will get biased.
I'm not as familiar with the MRF148, but I think it's 1/3 the device of the 150, so bias that between 15 and 500mA. My own experience suggests about 30 - 50mA for the 148 and 100 - 300mA for the 150.
The linearity requirements for MRI are not as severe as they are with SSB. So, even after getting biased, you may not like the sound of your signal. Ultimately, the best bias for SSB is whatever produces a clean signal. What I've given you is based merely on device ruggedness and survivability.
Although some manufacturers run the MRF148 and MRF150 as high as 65 volts, don't do it with a ham amp. The device breaks down at about 130V (per testing - the spec is actually more like 110, if I remember right). It's OK to run at the max if you're guaranteed a 1:1 VSWR. But if the VSWR rises due to a load that's more than 50 ohms, then the RF voltage swing increases, and the RF peak will damage the device. At 50V on the drain, you should be OK into 4:1. We ran the MRF150 at 48V and could run them into a short or open circuit without damage, although that's a bit outside actual spec for the device.
That's about all I can remember - it's been a few years. I always had the desire to make a ham amp out of our MRF 150 rigs - one of them was actually 20kW worth of FETs, with about 5kW worth of power supply (pulse amp - duty cycle allowed for lower power supply). But I never did it. And that 20kW amp covered up to 2 meters!
* assuming 8 devices
Copyright © 1998-2009 Dave Haupt W8NF. All rights reserved.
Published by kind permission of Dave Haupt W8NF.
Editing and page creation: A. Farson VA7OJ/AB4OJ. Last updated: 10/01/2018