IRF510 FET bias

[Wendymott]

Hi Peeps. I am currently trying to build a Top band AM tx....... I have tried various arrangements...... and at present have ended up with a single IRF510, driven by a BFY53...... and a DDS signal source.
The AM Modulator..... I have reverted to an Auto transformer, type as I was losing too much PA voltage in series mode.
The problem I have is.....Gate bias. There are various circuits available, with "0" bias.. i.e a 10K and a diode to chassis in parallel, and others with a bias pot from a stabilised supply, to set a standing current.
I initially set the bias pot for a quiescent current of 10m/A , with that I can get 4W RF into a 50R load........ well it starts off at 5W and drops down to 4W after a "warm up" period, I increased the heatsink and added a miniature fan and that slows the decrease in power, but it finally settles at 4W.
Where do I set the bias ? 10m/A or more or even get rid of the set bias pot.
The Drain current at 5W = 1 A @ 14V DC.....dropping to 800m/A

I wound a Double "E" transformer, Bifilar..... with 50T + 25T + 25T + 25T +25T ...... intending of finding the correct match between the RF and Audio driver...However...... the transformer core sings like a bird....driving the modulator input with a 1Khz sinewave. The cores are taped together....should I be gapping the cores to negate the DC current flowing.
Thats enough for now.... Im sure there are members out there metaphorically shouting at the screen
Your comment are most welcome

[Julesomega]

A commendable project. Pat Hawker in TT for Nov 1992 showed a circuit for a 7MHz "KISS" transceiver credited to G3RJT. I expect it's on the web somewhere, or I could scan mine

[Julesomega]

TT for March 1991 has a P-P "broadband" design for 1.8 - 7/14MHz giving 25W from a lower spec device, with the comment it should work better with IRF510. I'm not sure I understand his transformer details.

Since I stripped down some CFL light units I have been looking for an ultimately simple design to use the retrieved MOSFETs for the lowfer bands. The first pair I looked at has an in-house type number so I can't get a full spec, but the component tester reports a "capacitance" reading of 700pF. Can anyone suggest whether that would be Cg-s, Cd-s or Cd-g? Of course I could measure a 510 and see what the tester reports

[Julesomega]

The IRF530 datasheet shows 600pF for the "Input capacitance" which could be Cg-s + Cg-d under typical operating conditions. My component tester reports 1300pF without applied voltages.

All this is of no relevance to Wendy's original question as the RF amplifiers I have described are for CW use

Anyway, I'm still interested in the concept of RF use of switching style MOSFETs

[John_BS]

Wendy: if you have a look at the IRF510 datasheet, there are clues as to what's going on.

1. Although not set out in any detail, the gate-source voltage @ constant Id has a significant (but variable with Id) temperature coefficient.

2. Rds On similarly varies with junction temperature, rising from about 1 ohm @ 20C to 1.6 ohms @ 100C.

If you can't restore the 5W by tweaking the bias when hot, I'd suggest it's the latter which is causing the drop. Only solution would be to detect the mean output level and have an agc loop varying the raw HT to compensate.

You've not actually stated what your design aim is?

John

[Julesomega]

To continue my own questions, I have now actually read the datasheet and it shows unambiguously that the quoted Cinput is Cgs+Cgd. This is in perfect agreement with my component tester readings

[Wendymott]

Hi John. The design aim is a Topband AM transceiver with a DC input of 10W.. which was the original limit...... I decided not to tweek the bias until I had posted the above...there is still a bit to do, regarding the transformer turns ratio's.... that is why I gave myself a few variants... 25T = 1 layer of winding.
Regarding the bias.. I have a stabilising circuit with two 1N4000's in series with the 5V V reg ground pin, and thermally coupled to the heatsink... but not tried yet.
Ok Julian.... I will peruse the items mentioned in your reply

[G6Tanuki]

It sounds to me like you're not using enough drive: ideally you should feed it with a square-wave so the FET is driven fully-on, so minimising heat dissipation in times when the transistor isn't fully-on or fully-off.

Re-imagine the PA stage as a switched-mode power-supply rather than a semi-linear amplifier. I've seen several such circuits which use multiple parallel-strapped CMOS IC gates to drive the PA.

[Herald1360]

If you're driving the FET in class C, isn't zero bias OK? Just drive it hard enough to get it to take 10W from the drain supply. You'll need a few watts of audio to get a reasonable modulation level using a suitable transformer in the drain HT feed circuit.



That's my guess, anyway......

[G6Tanuki]

Here's an 'ideas' circuit from Pat Hawker's "Amateur Radio Techniques" which itself is an update on a circuit first published in Practical Wireless some time around 1980.

{have RF-capable power-FETs been around for 40 years? How time flies!}

Note the complete absence of bias to the PA, and square-wave drive [it uses a crystal oscillator at 2x the intended transmit-frequency, puts it through a divide-by-2 circuit to get a nicely-symmetrical 50:50 waveform, then paralleled CMOS gates to drive the FET]

[Wendymott]

Hi peeps.....thanks for your comments....yes Tanuki.... I have seen that before.... but I had a "blind side" as its not truly an RF circuit as we understood in the "old days".... I will give it some thought and look at it properly.

[John_BS]

Back-of-the-envelope calc suggests your observed drop is almost entirely down to the increase in RdsOn as the die heats up.

John

[cmjones01]

Originally Posted by G6Tanuki

Here's an 'ideas' circuit from Pat Hawker's "Amateur Radio Techniques" which itself is an update on a circuit first published in Practical Wireless some time around 1980.

{have RF-capable power-FETs been around for 40 years? How time flies!}

Note the complete absence of bias to the PA, and square-wave drive [it uses a crystal oscillator at 2x the intended transmit-frequency, puts it through a divide-by-2 circuit to get a nicely-symmetrical 50:50 waveform, then paralleled CMOS gates to drive the FET]

I've never read that article but I came up with the same idea independently: I built a transmitter for the university radio station when I was a student which used a TDA2050 power amplifier chip as the modulator with a MOSFET chopper on the output (I think it used a VN10 or something - whatever was available from Maplin, anyway), followed by filtering. It worked surprisingly well once I'd realised that it was important to run the power amp chip at less than half its maximum supply voltage to avoid overstressing it. This was in about 1993.

Chris

[G6Tanuki]

BTW the original circuit [which spawned the Pat Hawker revised version I posted] was in Practical Wireless July 1979 - available at the americanradiohistory site.

[Wendymott]

Hi John BS..... Thats exactly the result I was getting...... really the heatsink was not getting warm enough to start affecting the diode operation.

Ok Tanuki..... Im onto that thanks

Ok Chris. Im using a TDA2060 at present with 11W RMS into 4 ohms..and a +24V single supply..

I had a problem with my main "Puter" which was taking up "radio " time.... but its ok now...... so back to it... after my Covid jab weds

[Julesomega]

I knew I'd seen a really simple MOSFET PA design...
It was for 73+136kHz. I wonder how high that could be extended, 500kHz maybe?
The full article is at g3nyk.ham-radio-op.net

I see also an article by G0MRF on a P-P 250+W for LF/MF at g0mrf.com/lf.htm

[SiriusHardware]

Attached, the transmitter section from a relatively recent all mode 27/28 MHz transceiver - although the transmitter uses 2 x IRF520 in parallel to produce ~30W the bias arrangement for each FET is quite clear to see. The links J3 / J4 at the bottom are used to disable one output FET by turning off its bias while the bias for the other is being set.

[Julesomega]

Now you're talking - the circuit is genuinely aperiodic, I'm sure it will work all the way down to 7MHz!

[SiriusHardware]

There is some band - specific filtering on the RF output which I omitted from the 'section' I cut out because including it would have made the transmitter section about half the size and difficult to read by the time it had been pummelled by the forum engine.

[Wendymott]

I already have a 7 part low pass filter on the o/p...there are various coupling circuits around... I used a Pi arrangement.. as I dont have any HEF4069's in stock to try that other drive circuit...... it will wait until they arrive on saturday...