Home Brew RF dummy load showing reactance at higher frequencies

[MotoMike]

I searched for dummy loads and found nothing on point. If there is and I missed it I apologize. If this is not correct section let me know and I will re-post.

We need a high power dummy load in the 4kw range. for the amateur hf bands. The oil cooled "cantenna" type are at the most 1.5 kw. I have acquired 4 of the resistors used in these 1.5kw cantennas. my thought was to put two in parallel and two more in parallel then put those two parallel circuits in series to create a 50 ohm load. the resistors coming from a dummy load application are non inductive. I've seen home brew dummy loads built using parallel resistance to up the watt handling with good results.

My completed dummy load displays 52 ohms of resistance. when put on the hf bands, it is good at the lower ones but 4.5:1 in the 10 meter band, in the 28MHZ band. the construction is done on a piece of sched 80 pvc pipe with aluminum heat sinks under each mounting clip. the wire used is 12 gauge stranded copper thhn. crimped lugs were used and then all connections were additionally soldered. the wire lengths could be a bit shorter but are not excessively long.

Am I correct in thinking that with increased impedance I'm seeing at higher frequencies is inductance from something? I am anxious to see your thoughts.

 

[davenn]

......

Am I correct in thinking that with increased impedance at higher frequencies I'm seeing inductance from something? I am anxious to see your thoughts.

The wire-wound resistors you have used are going to increasing inductance as the freq is increased. This is normal.
Most guys will use a high voltage capacitor in parallel with the resistor network to null out the inductance.
Would start with inputting lower power and using a variable cap so that it could be tuned. once a good match is found
the v.c. can be removed and measured and replaced with the appropriate value fixed capacitorI am concerned that you intend to put anything over around 500W - 1000W through a PL259/SO239 plug socket.
That wouldn't be wise. You need decent RF connectors rated for the high voltage, current and kW rangeDave

 

[MotoMike]

Hi Dave
thanks for getting back to me so quick. The resistors are not wire wound, but are ceramic. They are made for oil cooled dummy loads.

My thought was that maybe making my connections with copper strapping as short as possible, rather than 12 ga stranded wire would reduce my inductance. I like the idea of using a cap to null the inductance though. I might get by without having to do a total rebuild. your concerns on the so239 are heard and considered. The duty cycle in the present application will be about 10 seconds then about a minute before the next transmit.

 

[davenn]

The resistors are not wire wound, but are ceramic.

you really sure of that ? ... .ceramic doesn't conduct
I would expect, and these looks like a coil on the ceramic tube
show me sharp up close and easy to read photos, specially of any info markings

My thought was that maybe making my connections with copper strapping as short as possible, rather than 12 ga stranded wire would reduce my inductance.

at 28 MHz, highly unlikely to be an issue, any lengths of wire in there would be such a tiny portion of a wavelength

. your concerns on the so239 are heard and considered. The duty cycle in the present application will be about 10 seconds then about a minute before the next transmit.

the PL/SO connectors are about the crappiest RF connector on the planet

I would be looking at high voltage N-type on the amplifier output and on any other assoc. gear hung on the output ... dummy loads, power meters etcDave

 

[berkeman]

you really sure of that ? ... .ceramic doesn't conduct
I would expect, and these looks like a coil on the ceramic tube

Agreed. A close-up photo would help.

@MotoMike -- do you have access to a low-cost hand-held LCR meter? Or maybe one of your other local HAMs has one? If so, you can measure the inductance of the power resistor, and that will tell you pretty quickly it if uses wire-wound construction.

 

[MotoMike]

Thanks for the help guys. I'm quite sure the resistors are not wire wound. I know that ceramic does not conduct, I think from looking at the construction that it is probably a doped mixture that has been fired. some quantity of carbon mixed with the slip to make 50 ohms. there is no evidence of wire on the resistor. I'll take a better shot for you when I get back in the shop on Wednesday.

they are of a similar construction to the one I've attached an image of here

.https://www.kanthal.com/globalassets/kanthal-global/downloads/resistors-and-capacitors/1-aa-series-800-1000-tubular-resistors.pdf

by my calculations the voltage I'll be dealing with is about 450. even the cheapest SO/PLs are tested at voltages well above this. my experience is that the failure will happen with cable heating rather than connector failure. Most manufacturers will test these connectors at least 3kv. I don't deny that So and Pl are lousy connectors, I just think they are up to the limited duty cycle they will experience here.

 

[sophiecentaur]

Would start with inputting lower power and using a variable cap so that it could be tuned. once a good match is found

Ideally, you would Bridge the resistors and find their inductance at low frequency. If you don't have that facility, you could get a value of its low frequency L by resonating it with a selection of C's from the 'drawer' ( a Marconi Q meter used to do this job pretty well). Then do a frequency sweep of the resistor and try to tune it with a variable / or various capacitors - starting perhaps with a C value that assumes no parasitic C and then reducing the value. But basically you should be working 'on the bench' first so that you will have an idea of what's actually needed. If you have the luxury of being able to 'look inside the box' then make use of it. If it transpires that the resistors are actually wire wound then you could easily find a ball park figure for the inductance with one of the many on line calculators.
It is much easier to find a solution tho this sort of problem if you can isolate the factors one by one. (Sorry if this is telling you how to suck eggs but it could save you a lot of time if you do it that way.) You may be handicapped by lack of test equipment - then you just have to be more inventive.

 

[MotoMike]

Ideally, you would Bridge the resistors and find their inductance at low frequency. If you don't have that facility, you could get a value of its low frequency L by resonating it with a selection of C's from the 'drawer' ( a Marconi Q meter used to do this job pretty well). Then do a frequency sweep of the resistor and try to tune it with a variable / or various capacitors - starting perhaps with a C value that assumes no parasitic C and then reducing the value. But basically you should be working 'on the bench' first so that you will have an idea of what's actually needed. If you have the luxury of being able to 'look inside the box' then make use of it. If it transpires that the resistors are actually wire wound then you could easily find a ball park figure for the inductance with one of the many on line calculators.
It is much easier to find a solution tho this sort of problem if you can isolate the factors one by one. (Sorry if this is telling you how to suck eggs but it could save you a lot of time if you do it that way.) You may be handicapped by lack of test equipment - then you just have to be more inventive.

Thanks for the help. The analyzer I have may be able to read the inductance of the resistors. I won't know till tomorrow when I return to the shop.

 

[tech99]

Thanks for the help. The analyzer I have may be able to read the inductance of the resistors. I won't know till tomorrow when I return to the shop.

If the resistor has inductance, then if you resonate it with a shunt capacitor, then the resistance value will increase.
I notice that the connecting leads are very long, by the way, which will add inductance. Why do you make everything so spread out?
I also thought the Cantenna was oil filled, but I presume I am mistaken. For a liquid filled load, if the power is applied for a short time only, much smaller resistors can be used.

 

[MotoMike]

If the resistor has inductance, then if you resonate it with a shunt capacitor, then the resistance value will increase.
I notice that the connecting leads are very long, by the way, which will add inductance. Why do you make everything so spread out?
I also thought the Cantenna was oil filled, but I presume I am mistaken. For a liquid filled load, if the power is applied for a short time only, much smaller resistors can be used.

Thanks for your help tech99. by saying that if I resonate with a shunt capacitor, then the resistance will go up. Do you mean to say the impedance will go up? my thought was that the resistance will remain constant and the reactive components in the circuit will change. If my problem is inductance, then the xsub L will increase with frequency and a parallel capacitor, because xsub C will decrease with increased frequency, could reduce and when the correct value is found solve my problem over a certain frequency range.

Yes the Cantenna type dummy loads are oil filled. As noted in my op, these resistors are sourced from such a dummy load. they are designed to be used in oil. the dummy load they come from is rated at 1kw in the hf bands for 10 minutes with a gallon transformer oil. since the application I have is for 4kw, I spread things out to allow for more heat to be carried away by the oil. the assembly will fit in a 5 gallon bucket with about 4.5 gallons of transformer oil. . my thought was that heated oil will rise pulling cool oil from the bottom of the can. I'm hoping the heated and cool oil will cycle owed to convection from the heat in the resistors.

 

[marcusl]

450 V is rms, the peak voltage is 632.
Also if you’re expecting a lot of heat, a PVC substrate might be a poor choice (your hot resistors are mounted right on the plastic).

 

[tech99]

Thanks for your help tech99. by saying that if I resonate with a shunt capacitor, then the resistance will go up. Do you mean to say the impedance will go up? my thought was that the resistance will remain constant and the reactive components in the circuit will change.
.

If you have inductance in the resistor itself, and you add shunt capacitance, then it forms a L-match and you will measure zero reactance but much higher resistance. On the other hand, if it is the connecting wires adding inductance, if you now place C across the actual resistor, such that reactance becomes zero, then you will measure an input impedance which is a lower value of R with zero reactance.
The VSWR of 4.5 is very high and I think it is a result of your long, meandering connecting leads. I am not too concerned about the connector you are using because VSWR of connectors is not an issue up until about 1 GHz and the connector has a thick pin for handling large current.
Have you measured it in the oil filled can - it might be better as it simulates a transmission line?

 

[Baluncore]

Has the high frequency been specified? 10MHz, 30MHz, 100MHz?

I would first eliminate those flying leads by replacing them with a coaxial cable to reach deeper into the cooler transformer oil. Then drive two parallel chains of two series resistors, without the long leads.

Mount the cylindrical resistors in perforated metal coaxial tubes, with inner diameter calculated to give 100 ohm line impedance in oil. Then you connect the two tubes in parallel to the 50 ohm line. Make sure the resistor current can return all the way up the inside of the perforated tube to reach the coaxial braid or socket ground. That balances and screens all RF currents.

For best HF performance, the resistors at the ground end of the tubes can be in a tapered perforated tube to further reduce ground terminal inductance. Solder the ground end of the resistor to the narrow end of the perforated tube.

PL-259 / SO-239 are not suitable for RF power. They were made for use with screened audio cable that originally used 4 mm banana plugs.

 

[davenn]

Has the high frequency been specified? 10MHz, 30MHz, 100MHz?

Yes, in the very first post ... the inductance was becoming a problem at 28 MHz ... amateur 10m band

PL-259 / SO-239 are not suitable for RF power

Yes, I have also made that comment way back in post #2 ... but the OP didn't want to take that advice

They are the crappiest connector on the planet and he should be using HV N-type connectors or similar

it would be VERY silly to use anything else ( other than other serious high power connectors) for a TX capable of up to 4kW

Dave

 

[MotoMike]

If you have inductance in the resistor itself, and you add shunt capacitance, then it forms a L-match and you will measure zero reactance but much higher resistance. On the other hand, if it is the connecting wires adding inductance, if you now place C across the actual resistor, such that reactance becomes zero, then you will measure an input impedance which is a lower value of R with zero reactance.
The VSWR of 4.5 is very high and I think it is a result of your long, meandering connecting leads. I am not too concerned about the connector you are using because VSWR of connectors is not an issue up until about 1 GHz and the connector has a thick pin for handling large current.
Have you measured it in the oil filled can - it might be better as it simulates a transmission line?

thanks. I'll work on it.

 

[MotoMike]

450 V is rms, the peak voltage is 632.
Also if you’re expecting a lot of heat, a PVC substrate might be a poor choice (your hot resistors are mounted right on the plastic).

Tech99
thanks again.
you're concern for the peak voltage is noted. You might be right on the PVC schedule 80 pipe. to say the hot resistors are mounted right on the plastic is not exactly true in that the mounting clips for the resistors are mounted on aluminum bars. my hope is that aluminum as heat sinks to help spread and dissipate the heat in oil will keep it from being a problem. If it is a problem, I'll change it. as noted the duty cycle will be about 10 seconds with about a minute before the next cycle.

 

[MotoMike]

Yes, I have also made that comment way back in post #2 ... but the OP didn't want to take that advice Dave

Dave, I thanked you for your input and your advice on the parallel cap confirmed my thoughts. to say I did not want to follow your advice is assuming facts not entered into evidence. My desires are not the only thing that figures in the project. silly as that may sound. Thanks again.

 

[MotoMike]

the help here is much appreciated. the analyzer I spoke of did read the circuit and confirmed that there was about 305nh inductance seen at the input. I used a variable cap in parallel with the resistive network and found I could null out the inductance at about 100pf I the mid 28mhz range. it allows the load to exhibit less than 2:1 swr 160 through 6. we have some door knob transmitting caps in the parts drawer that I am hoping oneis near the range I need. I will let you all know how it works out. the advice here was very useful.

kind regards,
Mike

 

[Baluncore]

the analyzer I spoke of did read the circuit and confirmed that there was about 305nh inductance seen at the input.

That 305nh inductance is equivalent of a 100 mm diameter single turn loop. That is fully explained by the flying leads you have. Replace those leads with current return straps that run parallel with the resistors like a transmission line, it will neutralise that reactance over a wide band. If you add a capacitor you will only make the load frequency dependent which will cause all sorts of problems.

 

[davenn]

to say I did not want to follow your advice is assuming facts not entered into evidence

it was in reference to you dismissing the connector advice, NOT the parallel cap advice

Again as I and Baluncore have stated ... you would be VERY unwise to use PL/SO type connectors at these power and voltage levels
They are total crap and that is a compliment to them

You came here asking for advice, you have been given it by people that have had many years of RF
work, should you choose to ignore it ... c'est la vie

I'm outta here

 

[MotoMike]

it was in reference to you dismissing the connector advice, NOT the parallel cap advice

Again as I and Baluncore have stated ... you would be VERY unwise to use PL/SO type connectors at these power and voltage levels
They are total crap and that is a compliment to them

You came here asking for advice, you have been given it by people that have had many years of RF
work, should you choose to ignore it ... c'est la vie

I'm outta here

Thank you Davenn

 

[MotoMike]

it was in reference to you dismissing the connector advice, NOT the parallel cap advice

Again as I and Baluncore have stated ... you would be VERY unwise to use PL/SO type connectors at these power and voltage levels
They are total crap and that is a compliment to them

You came here asking for advice, you have been given it by people that have had many years of RF
work, should you choose to ignore it ... c'est la vie

I'm outta here

Thank you Davenn.

 

[MotoMike]

That 305nh inductance is equivalent of a 100 mm diameter single turn loop. That is fully explained by the flying leads you have. Replace those leads with current return straps that run parallel with the resistors like a transmission line, it will neutralise that reactance over a wide band. If you add a capacitor you will only make the load frequency dependent which will cause all sorts of problems.

thanks Baluncore
I will round up the parts to make the arrangement more clean and shortest possible runs while trying to leave enough space for cooling. I have some copper sheet 1/16 inch thick that I can use to make the connections. It appears I may

have access to some high temp CE rated phenolic resin flat stock that would be better suited as the mount in surface. attached is my planned design.

 

[Baluncore]

. attached is my planned design.

You can further reduce the area of your RF current loops. Consider evolving the layout to further lower inductance; Label the resistors from the left, Ra, Rb, Rc to Rd on the right.

Connect the top of Rb & Rc to core of coax. Connect the top of Ra & Rd to braid. Tie bottom of Ra & Rb together. Independently tie the bottom of Rc & Rd together. You then get two smaller loops, but in parallel, so getting close to one quarter the original inductance.

But Ra & Rb form part of one RF return loop, so move them closer together to reduce the loop area. Do the same for Rc & Rd, making two separate loops, each with less area. That gets better than one quarter of original inductance.

Now consider mounting Ra on the front surface with Rb, right behind on the back side of the support structure. Do the same for Rc on the front with Rd behind. That gives them independent cooling flows and minimises RF loop area. As you have crossed the feed from the coax core to Rb&Rc, one on the front, one on the back, the loops are crossed and will cancel, and you have minimised the inductance. Make the top connections to the 50 ohm coax with twisted pairs, which will look like parallel balanced 100 ohm lines.

 

[MotoMike]

Thanks Balancore for that input. I had considered mounting them on opposite sides as you describe, but put them on one side hoping to create oil current up over the top cut out and down the other side to bottom cut out and up again. the thought that the 4 resistors be a steady force for flow. As I consider it the mounting on opposite sides would as you say allow for very short connecting runs, and they would be essentially mirror images of one another. 3rd proposed version attached. I could then get the resistors quite close together. As mentioned the on/off duty cycle will be in the 1:6 range and I'm using slightly more than a gallon of oil for each resistor, so more than the cantenna type load from which the resistors are sourced. As such, I'm thinking the cooling issue is not too significant. I do appreciate your help. attached.