New 1200 watt solid state MRF-150 linear - first light

[WA1GFZ]

The input splitter isolates the two inputs by at least 20 dB. same with the combiner. The fact that the VSWR is 2:1 between the amps and combiner i would look at the combiner and see  why there is a poor match

[Opcom]

I notice those FETs in parallel. Is there a 12-14V version for us mobile AM idiots? All the bigger stuff these days seems to be going 24-48V. I'd like to see about replacing the reworked old 8-transistor CB leenyar with something better as it's really poor below 7 MHz due to the "iron" and it would be better to try building a new amp than get into impossible mods on that dirty old hoar. It makes a 200W carrier with 100% modulation all day, but trying to launch that is a challenge. I have not built a solid state leenyar yet.

[WA1GFZ]

MRF150s won't run much power at 12 volts they really want at least 20 volts. Mil guys run them at 28 VDC. At 28VDC a pair will do 100 watts or so.

[W1VD]

T

Couple thoughts...if you haven't already found the problem.

Since these are *in phase* splitters and combiners, the two amplifiers need to be nearly identical to minimize the phase difference. Hard to tell from the pix but I assume the amps are built with identical parts. The silver mica compensating capacitors on the individual amplifier output transformers appear to be different - maybe just a different manufacturer - assume they are the same value.

One test would be to apply power to both amplifiers from the splitter and run each amplifier into it's own dummy load. Use a dual trace scope (with identical probes) to verify that the two amps are actually in phase. If they're not the problem is either the splitter or the amps. This can be determined by temporarily replacing the splitter with a resistive divider. If the two signals were in phase then the problem may be the combiner.

Other things to try if all of the above checks good.

Remove the ferrite beads cable tied to the transmission lines.

Attach the splitter and combiner modules (ground) directly to the common amplifier chassis.

Check to make sure that the ferrite material is the correct mix. Wind a few turns on the cores, measure the inductance and calculate the A sub L. 

It doesn't appear that you're using the 25 ohm resistors. Wonder how this affects balance? I use the resistors on my homebrew Granberg combiner for 137/500 kHz kW.

Do you have the CCI 4 kW combiner? Will be interested in what you find. 
     

[K1JJ]

Hi Jay,

It's still about  the same. I have it in use on the air and working FB at limited power, at < 300- 400w.

I'm using the 1KW combiner.

Right now the combined amp is using 40V and it puts out about 500W MAX  with the maximum of 20w drive or so.  It shud be doing at least 1KW out. The Fets stay luke warm, FB

What 25 ohms resistors do you mean?  Do they go on the combiner board? These combiners (splitters) were assembled by CCI, so I didn't add anything to them.

The boards are as identical as they shud be - at least they appear that way using original CCI parts.  I swapped the input and output combiner/splitter around with the same results. (They are interchangable)   I have the combiner and splitter bolted directly to the big heatsink/ common ground for the amps.

I measured the input attenuator that goes after the TX and before the splitter and it's 44 ohms J0.   However, the output to the combiner from each module is 2.5:1 swr.  That seems odd.  

I need to find a good second dummy load to run any phase tests.

In your comments above you said to remove the ferrite beads. Do you mean to try the amp with NO ferrite beads at all on the coaxes?  Or is that part of the phase measurement procedure above?  If so, I've not tried that yet.

At this point it's all buttoned up and being used as a clean class A-ish driver, so I don't need that much power outa it. At 40V it's running the cleanest IMD and it barely gets warm. So it does the job as intended.  But if I wanted to run it as a stand-alone linear, it would be lacking. Eventually I'd like to find the problem.


Side note: Wow, looks like the WX, OM has cleared up for Fri anyway.  HUZman is coming up here on Thurs and we'll be there early Fri. Rumor has it he may show up on the DX band using a "new" stack.  Hope to see ya up there Jay.

Later -

T

[WA1GFZ]

Jay might have a thought. remove the sleeves from the two outputs. Maybe it is effecting the combiner.  The sleeves force a balance at the amplifier but may mess up the combiner. ERB only has a sleeve after the last combiner. Still I can't see how it could give you a 2:1 VSWR. I think the transformer is messed up. Did you look at it with a scope? input to output of the combiner should show a voltage step up of 1:1.4.  I would think the sleeves can stay if the shields are grounded at both ends unless there is a ground loop around them with RF going through the chassis. The link I sent you showed common mode chokes with a lot more L. 8 MRF150s should make an easy 1200 watts but with a 1:1 transformer in the combiner maybe 1/2 that. Take a close up picture of the combiner transformer connections and send it to me.

[W1VD]

T

See drawing below. CCI shows the resistors in their drawings but says you don't need them unless you want the backup 1/2 power from one amplifier if the other amp STB. Dunno...have always used the resistors here.

Yes try removing the extra sleeves on the input/output coax lines.

Note that the splitter/combiner has it's own sleeves. These sleeves develop the isolation between the two input ports. The transformer then matches the combined impedance to 50 ohms. The sleeves do not dissipate any power so small ferrite cores can be used here and they should stay cool. The transformer section will have losses. Hard to tell from the pix but the transformer core looks small for the kW power level.

Guess you have an impedance analyzer of some sort. Check the splitter and combiner out of the circuit. Terminate both input terminals of the combiner with 50 ohm resistors (disconnect the amp) and look at the impedance at the combined port. The VSWR should be very good. I typically see >20 dB return loss (and isolation >20 dB between ports)...at 137 and 500 kHz...

[K1JJ]

Jay,

So far removing the cores from  between the splitter/combiner to the modules had no effect on operation.  Right now at 40V  I'm doing 40A input (1600w) and getting out about 500w max on 40M. On 75M it's about 400w out.  That's about  25-31% eff,  piss poor..

I tired setting all the fets so they idle at the same heat but no effect as you'd expect.

I see a slight diff in current between the 2 modules. One will do 10A while the other is doing 15A.

I'll disconnect the combiner and try the 50 ohm test later.


Question:  If I can idle all the Fets and they all pretty much show the same gate voltage for 1/2A idle each, does this mean they are all OK?  ie, Can one be "soft" for RF drive just like a tube with a filament having poor emission?


Also, how exactly do I set the bias for each?  Right now I'm watching the total current for each module and progressively adding 1/2 A per fet until I see 2A total. Is this correct? The problem is some idle warmer than others after doing this.  I could set all the gate voltages the same, like 3.5v, but that seems arbitrary when they idle at different currents..

T

[WA1GFZ]

Good, For the life of me I could not figure out how a common mode choke would effect anything but still a good test. Tom I would set the current on one FET then wait a while  and see if it drifts up a bit. Then move on to the next one. Since the FETs are not matched they will have slightly different transfer curve. Look at the app note I sent you on the combiner. gain goes up with increased bias and efficiency goes down. Try setting the bias per the EB104 app note as a check.

[W1VD]

T

10A vs. 15A is more than a slight discrepancy! Within an amp would be more like it. Do they exhibit the same idling current?

If you're in the mood to make additional measurements...

* While the splitter and combiner are out test each amplifier module for power output using exactly the same input power.

* Check the splitter and combiner for balance. Apply a small amount of power to the combination port and look at the output ports on the dual trace scope. They should be very close in amplitude and phase. Make sure to terminate  both outputs with 50 ohms for the test.

Idling current is more important than gate voltage. Your adjustment procedure should be fine.

The first test above should answer whether one amp is soft or not. Not sure how much variation there is from batch to batch. They do offer matched sets, though. 

       

[WA1GFZ]

What Jay said.. bias does change with heat.  50% offset is a big deal.

[Mike/W8BAC]

Hi Tom,

I think it was asked earlier, I keep looking at your efficiency numbers and think "this power is being dissipated someplace". Is it possible you have a spike on VHF? It doesn't make sense. I'm sure you have a spectrum analyzer in the shack. Maybe that watt meter is only showing RF in the HF range and you might have a spike up high when on 75 meters. Just my thoughts. Best of luck.

Mike

[WBear2GCR]

Nothing that will help with the present problem, but long term given the heat that will be there, the copper on aluminum is surely going to make for nice electrolytic corrosion...

But is the excess power showing up as heat??

                _-_-bear

[K1JJ]

I've made a lot of measurements and tests.  Got some more to do yet. Here's what I have so far:

The two combiner splitters measure identically.  Here's the test:

I put 50 ohms across the two inputs and measured the output with the MFJ-259B. I get 68 ohms j0  -- on both of the units.
I then terminated the output with 50 ohms and one of the inputs with 50 ohms and measured the other input. It shows 78 ohms j0.  I then put the 50 ohms on the other input and measured the opposite input and saw 78 ohms J0 again, as expected.
Hopefully this will tell if my combiner/splitter is OK in ratio, etc.


I keyed the amp and measured the input impedance of one module and see 40 ohms j0.

I have both amps running separately and for the first module (w/o combiner and splitter) I see on 75M:

With 2.5w in I get out 150w = about 18db gain.

This is 15A at 40V = 600w input power.  150w/600= 25% eff, poor.  I don't feel anything getting hot, so not sure where the other 450w is going.  I checked my current meter against the VOM and they are close.  I'm using a dead carrier to measure  power out and current.  I measured the RF input using both the scope and the meter on the FT-1000D showing about 2.5w of drive. The output power is measured using the Bird wattmeter . I see the FT-1000D puts out its normal 200w on the same meter and coax line, so that's accurate.

Gonna measure the 2nd module to see if they match up and check it on 40M also.

The math says if I drive one module with 7.5w, which is the maximum, I will get out about 450w, but the current wud be 30A! That wud equal an input power of 1800W, which wud smoke the Fets, being only rated at 600w dis.  Something isn't making sense here.


More measurements to come...
T

[K1JJ]

I measured both modules individually.

I found one of the current meter wire shunts had come loose, so now both modules are close in total current.

Module #1 measures 2.5w in and 125w out at 14A /40V = 125w/560w in = 22% eff.
Module #2 measures 2.5w in and 120w out at 13A/40V =  120/520w in = 23% eff.

It appears the problem lies with the module's efficiency. Maybe I'm measuring something wrong or making an incorrect assumption in power calculations?  The output transformers run only luke warm at most as a long carrier. I see no other heat problems and the Fets are only warm to the touch after a carrier.  I don't get it why the eff is so low when it shud be up around 60% or at least 50% using 40V.

I found the results to be similar on 40M, just slightly lower power output as expected.  This is with no LP filters.

Next I will combine the two again and double the input power to 5 watts and see if I get out the expected 250 w.


UPDATE:   I measured both modules individually again and see they are almost the same in performance and efficiency.  125w out at 560W in =24%.  13A at 40V.... both . On 40M I see an extra 25w out at the same parameters, almost the same.

I will combine them now.


This is strange cuz I thought the first module I built a year ago had decent eff and worked up to 600w out. Now I am having troubkle seeing 300w out without exceeding current in.

T

[WA1GFZ]

Tom,
Could it be possible your meter shunts are warming up screwing the accuracy at high current. A copper shunt will increase resistance about .4% per degree C. at 40 amps each shunt should be at least #12 but dual #12 would be more stable.
You also need to wire the meter properly as a 4 terminal measurement. Copper flashing makes a goos shunt mounted right on the meter screws. Do you suffer from JS?
It doesn't take much JS at 20 amps to really mess you up.
I thought that power supply only put out 35 amps but don't know where it folds back because never ran it that hard. Your heat sink is pretty small so  if you were burning that kind of power you could fry eggs on it. Since both amps are doing the same thing and Combiner seems fairly close maybe it is your metering

[WD5JKO]

Module #1 measures 2.5w in and 125w out at 14A /40V = 125w/560w in = 22% eff.
Module #2 measures 2.5w in and 120w out at 13A/40V =  120/520w in = 23% eff.

    Tom, If these AMPS are class AB, isn't the % efficiency supposed to be around 60% at maximum unclipped output? If so, doesn't this efficiency drop with reduced drive? So if those AMPS you are using are designed for 600 watts at about 60% efficiency, doesn't the efficiency drop in half when we reduce the drive such that the output is 1/4 (-6db) of maximum.

   Think of this like an AM linear that when everything is optimized will have an efficiency of 33% with headroom for 100% modulation. In this case we also back off the drive 6 db (say from 100W to 25W).

   Maybe your data isn't too far off at reduced drive conditions.

Your heat sink is pretty small so  if you were burning that kind of power you could fry eggs on it.

   Maybe so, but I wonder how long the carrier was running. Maybe the heatsink was just starting to get cooking when the carrier was shut off.

Jim
WD5JKO

[K1JJ]

Jim,

That's exactly what I was thinking - maybe at reduced power the eff is poor.  Yes, it's that way with my tube Am linears, etc. Though, even when running ssb and I'm hitting it as hard as I dare, I see only about 500w using BOTH modules. I shud easily see 1KW on the peak reading wattmeter.  I'm not done testing yet tho.

Frank - I ran the last few power tests with the VOM ammmeter in line and recalibrated the amp meters. They are now with 1/2 A accuracy and the measurments reflect this.  I have run 200 w carriers for up to a minute without excessive heating on that heatsink. On ssb, the fan keeps it pretty cool.

It appears from logic that the "problem" if any, lies in the modules, not the combiner/splitter.  (Frank, you thot the data looked good for them) 

I still plan to combine them now and see if eveything double when I drive it with X2 power.

The PS will do 40A and I see the voltage stays steady. I hear some hum-buzz from it, but that's normal... :-)

T

[K1JJ]

Well, after all this testing I think I may have stumbled across the problem.

Firstly, the combined modules are working as they should. I'm getting about double the power out with X2 the drive.

Here's what I found wrong:
Usually, when I try a strap test, I look closely at the FET's current to be sure I don't go too far. Once in a big while I would hear a small squeal, but thought it was just feedback on the board or maybe some arcing.

Anyway, this time I tried to see what caused it. As it turned out, once I go past a certain limit like, 35A or so, the 40V switching power supply starts to fold back.   As I goosed the drive, I saw the digital meter drop back all the way to 18V!!  I usually missed it cuz the digital VOM is too slow to react. An analog meter wuda showed it, or so wud the scope.

I just hooked up the scope and watched the DC. When I get close to 36A, I see some funky ripple and then it squeals and collapses.   Maybe some RF getting in too. I will try some more beads on the supply leads, though I have some there already.  Frank, you mentioned the supply was good for only 35A or so, so right on!

I'm hearing kinda a raspy buzzing sound coming outa the PS, even at the 250 watt output level. I see some RF on the top of the DC waveform too. It gets worse as I increase power until it collapses.  Well, at least this is progress.

Frank, do you think a bank of bear cans will help?  I have about 500,000 ufd left over from the PDM rig I cud strap in there.  At least it wud help for peak power on ssb.

BTW, I saw about 700w peak out on 75M before it folded back, so maybe we're OK with the amp.

If it doesn't work out, this will be OK as-is as a class A driver, it's intended purpose and as a stand-alone linear when I can find a bigger supply.


We'll see what happens.

Thanks for the help guys. At least it appears the amp is doing what it shud.

Later -

T

[K1JJ]

With the amp operating all put together, I notice that the power is still down somewhat on 75M.  I can get a robust 750w pep out on 40M with little buzzing from the supply. On 160M it's about the same.  But on 75M I hear a louder buzzing earlier on and the power out is limited to about 550w  before the supply collapses.   It seems that maybe 75M is getting some crap into the modules?

It appears 75M has some kind of parasitic and making the supply situation worse at the same time.  I see the supply hold firm, but the output power of the 75m amp is just not there.  Though, from earlier measurments, all bands seem to put out 125w with 2.5 w drive, incl 75M. Maybe as the RF level gets higher the problem starts.  The PS may be susceptable to 75M RF?  Even tho it's staying at 40V, maybe it's injecting a parasitic into the amp. I can hear the switching freqs on 40M, FWIW.

Any suggestions of what to try? I will try the bear cans tmw if you say to, Frank/Jay.  I added a lot of ferrites at the supply, but NG. Might try the AC power cord tmw too.


T

[WD5JKO]

Tom,

   Is the suppression at the power supply output common mode or differential? I believe that if the power supply is mis-behaving due to RF, then you might need both types of suppression; kind of like a RF version of what we have in those metal boxed AC line filters.

    You could try four 12V lead acid batteries (say trolling motor marine batteries) in series, with a good fuse or breaker. 

Jim

[W1VD]

T

Good info.

First fix the power supply and metering problems. I've found many high current power supplies (even HP) do strange things with rf. Sometimes it's the actual power output that's changing but other times it's just the metering becomes inaccurate. Had one situation where the metering shows the voltage and current rising to unexpected levels while the output was actually steady. The cure was a .01 right at the output terminals and a ferrite choke. Twisting the + and - power supply leads (think twisted pair) between the power supply and the amplifier also seems to help.

The transformer ratio in the splitter/combiner is incorrect...if the MFJ is to be believed. It would appear that the tap point is off as the ratio is incorrect - about 1.65 instead of 1.41. You could try changing the tap point ...or just leave it as is. It should exhibit 50 ohms all around if working properly.

Once the power supply/metering issues are fixed the measurements will make more sense. 

   



       

[K1JJ]

Hi Jay,

Thanks for the input.

The metering is separate from the power supply and has been fixed. I'm using a pair of 3" round old buzzard meters with a shunt across each. They agree wid the VOM ammeter, even when the supply starts to fold back.

I finally looked at the PS tag and see it says 34A. That's why I'm folding back when I hit about 18A per module.

If you look back at my last 2 posts you will see the remaining problem is 75M seems to be running lower power than the other bands when I go past the 200w output point or so. The PS has a louder buzz coming from it. But the voltage stays constant at 40V. It's like I have some kind of parasitic that is limiting the 75M band to 500w or so into saturation causing poor efficiency at higher powers. 160 and 40M do not exhibit this limitation and appear like they want to do 1kw with the proper PS.

Do you think a bank of 500,000 ufd beer cans would help the supply?  Any other suggestions to calm down the 75M problem?

I will try a different tap point on the combiner too.


Hmmmm... wonder why both splitter/combiner boards wud be off?  One board is from 1992 and the other is recent. Seems odd. They are wired exactly as I see in pictures, by the factory.


Jim - I will try some of your PS suggestions. Though, the problem will still be I need 50A from a 34A supply. Might have to acquire or build a bigger one if I want to run the amp as a stand-alone. Otherwise, as a driver it is FB as-is.

T

[WA1GFZ]

Tom,
I would take a bunch of beads from the class e rig and string both positive and negative lead through them to make a big common mode choke. Put the .01s at the amp side of the choke. Make sure the chassis of the supply is grounded with a 3 wire cord. I think you are getting RF into the supply. It is only rated for 35 amps so that is about 1400 watts at 40 volts. The amplifier should be running around 50% efficiency maybe a tad more so 750 watts out is actually pretty good. Check the data sheets on the MRF150 this isn't a tube. You could run the voltage up a tad but the higher you go the less safety factor you have. I think the EB104 works best around 42 volts based on everything I read. If you want 1200 watts then you need at least 2500 watts of strappage feeding it. BTW I do have 2 more of those supplies.
Adding a big bank of caps is a double edge blade. It will help with peak currend but also provide lots of julies if there is a fault.
All good tests when I try to do 16 MRF150s

[K1JJ]

OK, Frank -

I'll add the torroids in the PS leads as you suggest and the .01.

OK on the beercans. That's why I axed to be sure. I may add one 100,000 ufd just to see the results, if any.

Tell ya what - bring that other PS to HossTraders this weekend and let's make a deal. Either cash or I'll bet I have some stuff here you can use. Need another one of those EQ's ? or I also have an Alesis SS dual band audio compressor I'm not using.  Both work FB... What you need?

I could put one PS to feed one and the other for the 2nd module.


OK on the 750W.  I think it's doing OK now too, just as long as I get that 75M fold back fixed. My guess it's PS related cuz I've checked about everything else.  Hopefully the slight mismatch in the combiner that Jay sees is not significant. I'll bet they are working as designed and if we tested some more from the factory, they'd measure out the same.

You're going to Hoss with the nice WX forecast, right?  Huz and I will be there early Fri.

T