Analyser wrote: I might have a go at building one of these amps for the guy that asked for a schematic; it's not really fair to put 2-3 versions of a theoretical design and expect someone with limited experience to build all of them and be a ginuea pig.

I think that would be really good - if you're in a position to do it. I had the same thought myself, but know I wouldn't be able to get round to it any time soon.
I don't think I ever tried a single "L" match, but used double "L" matches quite a bit; also Tee Section CCL match topology. Didn't often have this level of detail for the transistor specs though, so often used (Vo*Vo)/2Po as a rule of thumb starting point for the maths.
I'm wondering how much heat this output network is going to generate, but that may be unfounded.
Thanks for the additional info. I had a feeling that some sort of CAD was probably involved.

I got all the parts needed to build this. So there is a problem with the output matching network? Is it just a matter of changing L3 until it's best or do you think it needs to be a different type of matching network for it to work effectively? Should I start building it or wait?

OgreVorbis wrote:I got all the parts needed to build this. So there is a problem with the output matching network? Is it just a matter of changing L3 until it's best or do you think it needs to be a different type of matching network for it to work effectively? Should I start building it or wait?

Start building it but make L3 a hairpin (U shape inductor) of about 22mm long and bent over a 7mm drill bit. Experiment with the length by cutting 1-2mm off the ends and trying it again. If you hae time start with it longer , say 30mm and work down from there.

Analyser wrote:

OgreVorbis wrote:I got all the parts needed to build this. So there is a problem with the output matching network? Is it just a matter of changing L3 until it's best or do you think it needs to be a different type of matching network for it to work effectively? Should I start building it or wait?

Start building it but make L3 a hairpin (U shape inductor) of about 22mm long and bent over a 7mm drill bit. Experiment with the length by cutting 1-2mm off the ends and trying it again. If you hae time start with it longer , say 30mm and work down from there.

Yes. Apologies for throwing too many variables into the thread.

OK, I just finished the board, but I haven't soldered the transistor or mounted on a heatsink yet. What should I set the bias voltage to? I used 100pf on the two output matching (C5, C6) caps instead of 120pf because the 120pfs I have seem pretty weak.

So I've mounted the transistor. I have an antenna analyser and I hooked it up to both the input and output of the amp with no power applied. (I'm not sure if this is a valid why to check tuning. Maybe the amp has to be on first to see the matching?) Anyway, I found that having 3x 100pf on the output along with 2 turns on the output coil gives an SWR of 1.4 at 98MHz. The original proposed way gave about 3.0 SWR at 71MHz. I haven't figured out the input yet. I'm not even sure if this is the right way to do this, but I'm afraid to turn it on if the mismatch is too high.

OgreVorbis wrote:So I've mounted the transistor. I have an antenna analyser and I hooked it up to both the input and output of the amp with no power applied. (I'm not sure if this is a valid why to check tuning. Maybe the amp has to be on first to see the matching?) Anyway, I found that having 3x 100pf on the output along with 2 turns on the output coil gives an SWR of 1.4 at 98MHz. The original proposed way gave about 3.0 SWR at 71MHz. I haven't figured out the input yet. I'm not even sure if this is the right way to do this, but I'm afraid to turn it on if the mismatch is too high.

I guess that's a good way to start but the antenna analyser will provide too weak a signal to provide good tuning information. I would start off with zero bias and apply maybe 1 watt rf input and then start increasing supply voltage. If everything looks ok then set the supply to 12v and add more input. Stop at 5w. After all that is done you can start increasing bias but note this is not done in voltage but current when the amp and drive is turned off. If I remember correctly I saw a figure of 2A as a max bias current but that seems huge to me so personally I would stop at 500mA or 1A at a push.

I built a couple of amps using the (almost identical) circuit to the one you showed earlier in the thread. The output match was quite broad - gain flat for about 3MHz either side of the frequency it was nominally set up for - but not truly broadband. I found that with 4 Watts of drive, I could get about 60 Watts output, but the bias current setting was very critical. The power output would climb steadily as the bias was turned up until - suddenly - the power would jump from 30W to 60W. Increasing the bias beyond that point caused the power to start to droop. The critical current on my two example was about 450mA.

Interestingly, the bias point didn't seem to vary much with temperature of the FET. The procedure seems to be to turn it up slowly (I used a 5-turn preset) until you get the power jump, then advance it a tiny bit more.

All my initial testing was done at 12.5V (since that was the output voltage of the old "CB" PSU I used), but increasing the drain voltage seems to increase the gain. I get about 75 Watts out at 15V and 4 Watts of drive at 98MHz. Power output droops slightly as heatsink temperature rises, but always stays above 70 Watts.

I tried turning up the drive, but gain compression sets in at about 4.2 Watts of drive in this circuit.

I have a 7-pole output filter and a very clean driver. The second harmonic is in the grass on the analyser (better than -90dBC) and third is -85dBC (though that might have been coupling to the probe from the drain circuit due to my sloppy bench practice!)

@Albert
Good to have you back here on radionecks

Thanks, Head!

It's good to be back.

Well, I can't make it work. At best I got about 4.5W out with 1W input, but the SWR on the input is terrible and I can't improve it. Increasing the bias nothing happens and then it suddenly starts drawing full current. Same thing happens when I tune it sometimes. I unplugged it instantaneously and the voltage reg has an internal current limit so I don't think it's broken.

I don't think I have enough experience to continue messing with this unless someone gives me some concrete numbers.

Maybe look at this 70MHz amp and start from there
http://www.sp2dmb.cba.pl/70e/RF%20ampli ... 70HVF1.pdf
The output is almost the same as the one here on RN but the input coil and capacitors are different
So try to change the input like this and maybe use a coil with 1 turn less

Hi RF Head, the link you gave us just brings up this picture. ELo

This link should work. I think this is the one RF Head meant http://www.sp2dmb.cba.pl/70e/PA70H_eng.pdf ?

I agree that this is logical, as the input match seems problematic, but also wonder the following.

Ogrevorbis, were you monitoring the Gate voltage, as well as the Drain current. If so, did you notice the gate voltage where the FET started drawing heavy current ? The reason I ask: the RD70HVF1 should draw next to nothing, until the Vgs reaches about 2v. Over 2v the current consumption will rise rapidly (see Vgs-Id characteristics top right of the attached). So depending on Vgs at the time, it could be related to DC conditions; it looks like you probably have quite a wide sweep on the bias pot. On the other hand, it could also be the increasing gain - with increasing Vgs, causing it to take off. As you said it's also drawing high current sometimes when you tune it, that could also be symptomatic of some kind of self oscillation. Were you monitoring the output in any way - other than power measurement. Wondering if there was any indication of "nasties" when the current shot up. I know you freely admitted that you don't have a lot of experience, are you confident in the stuff like grounding and layout ?

Otherwise I guess just try what RF-Head suggested, and report back. Even if it is taking off, if the input match is as bad as your SWR measurements suggest, it could be causing or contributing. On the other hand, if something is causing the driver to take off, it could be trying to drive with nasties, which could also explain a bad SWR which doesn't respond to tuning.

EDIT: Unfortunately the units don't seem to show in the datasheet. The horizontal axis is Vgs in 0.5v increments. The Vertical axis is Ids in 1A units. Again, this is the graph at the top right of the page.

Built In A Shed:

That link doesn't work either!

Albert H wrote:Built In A Shed:

That link doesn't work either!

Strange. It works ok for me, but might need pdf plugin.
Attached (hopefully) is the schematic - including credits and ancillaries.
Welcome back Albert. It's good "hearing your voice" again !

Albert H wrote:Built In A Shed:

That link doesn't work either!

Working fine here.

Ps great to see you back sir.

Doesn't for me. Maybe it's just Firefox?

OgreVorbis wrote:Well, I can't make it work. At best I got about 4.5W out with 1W input, but the SWR on the input is terrible and I can't improve it. Increasing the bias nothing happens and then it suddenly starts drawing full current. Same thing happens when I tune it sometimes. I unplugged it instantaneously and the voltage reg has an internal current limit so I don't think it's broken.

I don't think I have enough experience to continue messing with this unless someone gives me some concrete numbers.

I know you're frustrated but try not to give up just yet! People on here will give you plenty of pointers. I would build an amp and try to get it working for you but I'm away from the workshop for a few weeks so can't offer much practical help at the moment.
In the meantime, as others have suggested, try increasing the input coil and decreasing the input capacitoance before the coil. I would try a 3 turn coil of 6mm diameter and a 65pf trimmer to ground. Back off the bias and start again with the tune up process I mentioned before.

What about some impedance transformer?


Sent from my iPhone using Tapatalk