I think it would be nice to set a target threshold in advance to establish if NiCr is worthwhile. i.e. if it has a creditable edge.
I would suggest the threshold should be if the gain of a typical amplifier at VHF is reduced by a relative reduction factor of about 1dB when you compare identically wound suppressors with the same resistor value. One suppressor is NiCr and one is TCWire.
In reality, people are going to claim you could simply stick with TCWire and just change the resistor slightly to make up the difference. This is very true but that isn't the issue here. The issue here is to see if NiCr has an edge or if it is 'voodoo snake oil'. Personally, I don't think it will quite fall into the latter category but if it beats the 1dB gain reduction on a typical amplifier model then I think it can claim some merit. I'm expecting a ballpark 5% reduction in Rp between the two as a guess at 100MHz. This is about the same as the tolerance of a typical resistor. If it performs better than this then it gets interesting. I can't see it getting anything like the same relative improvement as that suggested in the N7WS data because the tests weren't really a fully controlled comparison. You only need a small amount of stray reactance in the construction somewhere to add an error that will be significant. So the suppressors need to be as similar as possible.
I'd also add that I think the relative change in Rs has more meaning than looking at Rp because it's the Rs figure that ends up defining Rs in the model for the entire suppressor/htchoke/linkwire/Pi tank network seen by the anode as Rs, XL at 100MHz.
Note that to arrive at an amplfier gain reduction figure you have to include a model for the tank circuit at VHF that resonates at the problem frequency. To get a 1dB gain reduction the modelled anode load has to see a relative reduction of about 12% across the two suppressor types.
I'll do my best to ensure a fair comparison. I'll go as far as measuring/matching the resistors to minimise errors in comparison. I'm slightly worried about soldering the NiCr so I may simply wire wrap over it tight with the resistor leg and then solder it as Rich suggests. Any more tips for soldering (solder type?) will be useful. Also, what frequencies should the test be done at? 100MHz seems quite low for a single tube amp so maybe there should be two tests. 100MHz for a dual amplifier model and maybe 180MHz for a single amplifier model.
Is it easy to bend? I've never played with it before and the stuff I've bought is about 0.9mm diameter. I'm not the neatest constructor but as long as they look the same and I match the inductance then it's fair
OK the NiCr wire arrived today but I've only had a quick and informal play so far. I don't have the high power resistors so I've used 100R 0.6W 1% tolerance rated tiddlers. I made up two suppressors similar in dimension to the N7WS description. i.e. 4.5T on a 1/4" tool. I tried very hard to make both look the same and also measure similar inductance on a VNA. I've literally only had a quick look but as expected the NiCr has slightly lower Q. Now I don't think Rp is the the most intuitive thing to look at for a comparison but I got about a 9% difference at 100MHz and 6.5% difference at 180MHz. This is slightly more than I expected but not anything like the 40-50% reduction claim. However, a lot depends on symmetry of construction and also the amount of stray reactances in the construction. So I see the above result as unreliable at the moment.
However, as I said in the last post, this isn't as useful or intuitive as looking at the change in "series equvalent" Rs. So I fine tuned them to have the same series reactance at 100MHz as indicated on the VNA.
Basically I measured Rs as:
24 ohm series R and 49 ohm series inductive reactance for the NiCr suppressor at 100MHz.
21 ohm series R and 49 ohm series inductive reactance for the conventional TCW suppressor at 100MHz.
This is about a +14% change in Rs for NiCr and is much more interesting because this series resistance dominates the series R of the entire network at the anode. Note that the relative change gets slightly diluted by the contribution of the series R (ballpark a couple of ohms?) already in the tank network.
I must stress this isn't a definitive test yet. I need more time to try a few things but I do think NiCr is going to prove it has 'some' merit. I also think there could be ways to rearrange the suppressor system to better exploit the NiCr and maybe get a better improvement.
I did plug the two series equivalent networks into a very basic system model of an amplifier and the extra loop gain reduction from the NiCr suppressor was just under 1dB at 125MHz. So it appears to be doing something useful.
At this level of change it's definitely not voodoo snake oil, but then again it's not exactly a cure all for instability. Definitely worthy of a bit more testing though and I'll do some more formal tests over the weekend and post up a few VNA plots
OK the NiCr wire arrived today but I've only had a quick and informal play so far. I don't have the high power resistors so I've used 100R 0.6W 1% tolerance rated tiddlers. I made up two suppressors similar in dimension to the N7WS description. i.e. 4.5T on a 1/4" tool. I tried very hard to make both look the same and also measure similar inductance on a VNA. I've literally only had a quick look but as expected the NiCr has slightly lower Q. Now I don't think Rp is the the most intuitive thing to look at for a comparison but I got about a 9% difference at 100MHz and 6.5% difference at 180MHz. This is slightly more than I expected but not anything like the 40-50% reduction claim. However, a lot depends on symmetry of construction and also the amount of stray reactances in the construction. So I see the above result as unreliable at the moment.
However, as I said in the last post, this isn't as useful or intuitive as looking at the change in (series equivalent) Rs. So I fine tuned them to have the same series reactance at 100MHz as indicated on the VNA.
Basically I measured Rs as:
24 ohm series R and 49 ohm series inductive reactance for the NiCr suppressor at 100MHz.
21 ohm series R and 49 ohm series inductive reactance for the conventional TCW suppressor at 100MHz.
This is about a +14% change in Rs for NiCr and is much more interesting because this series resistance dominates the series R of the entire network at the anode. Note that the relative change gets slightly diluted by the contribution of the series R (ballpark a couple of ohms?) already in the tank network.
I must stress this isn't a definitive test yet. I need more time to try a few things but I do think NiCr is going to prove it has 'some' merit. I also think there could be ways to rearrange the suppressor system to better exploit the NiCr and maybe get a better improvement.
I did plug the two series equivalent networks into a very basic system model of an amplifier and the extra loop gain reduction from the NiCr suppressor was just under 1dB at 100-125MHz. So it appears to be doing something useful.
At this level of change it's definitely not voodoo snake oil, but then again it's not exactly a cure all for instability. Definitely worthy of a bit more testing though and I'll do some more formal tests over the weekend and post up a few VNA plots
OK the NiCr wire arrived today but I've only had a quick and informal play so far. I don't have the high power resistors so I've used 100R 0.6W 1% tolerance rated tiddlers. I made up two suppressors similar in dimension to the N7WS description. i.e. 4.5T on a 1/4" tool. I tried very hard to make both look the same and also measure similar inductance on a VNA. I've literally only had a quick look but as expected the NiCr has slightly lower Q. Now I don't think Rp is the the most intuitive thing to look at for a comparison but I got about a 9% difference at 100MHz and 6.5% difference at 180MHz. This is slightly more than I expected but not anything like the 40-50% reduction claim. However, a lot depends on symmetry of construction and also the amount of stray reactances in the construction. So I see the above result as unreliable at the moment.
However, as I said in the last post, this isn't as useful or intuitive as looking at the change in (series equivalent) Rs. So I fine tuned them to have the same series reactance at 100MHz as indicated on the VNA.
Basically I measured Rs as:
24 ohm series R and 49 ohm series inductive reactance for the NiCr suppressor at 100MHz.
21 ohm series R and 49 ohm series inductive reactance for the conventional TCW suppressor at 100MHz.
This is about a +14% change in Rs for NiCr and is much more interesting because this series resistance dominates the series R of the entire network at the anode. Note that the relative change gets slightly diluted by the contribution of the series R (ballpark a couple of ohms?) already in the tank network.
I must stress this isn't a definitive test yet. I need more time to try a few things but I do think NiCr is going to prove it has 'some' merit. I also think there could be ways to rearrange the suppressor system to better exploit the NiCr and maybe get a better improvement.
I did plug the two series equivalent networks into a very basic system model of an amplifier and the extra loop gain reduction from the NiCr suppressor was just under 1dB at 100-125MHz. So it appears to be doing something useful.
A practical VHF suppressor should be able to reduce the anode's VHF RL - and VHF amplification - by at least 3/4.
At this level of change it's definitely not voodoo snake oil, but then again it's not exactly a cure all for instability. Definitely worthy of a bit more testing though and I'll do some more formal tests over the weekend and post up a few VNA plots
I'm using 0.9mm diameter wire according to the spec sheet.
A practical VHF suppressor should be able to reduce the anode's VHF RL - and VHF amplification - by at least 3/4.
Yes it should reduce it a lot. However, the extra 1dB gain reduction target above is a relative figure when you swap a conventional suppressor for a similar NiCr one. It looks like this target will be met but I need more time to do some more tests that are more accurate. The previous tests were quick S11 tests on a VNA after a SOL calibration. I really want to measure it as a two port network so I can insert the 2 port model into a simulation.
None of my 2 port test sets here at home are big enough as I usually work in a world where an 0805 SMD resistor is a fairly large component. So I need to make something and make sure the residual test set errors are small.
This afternoon I spent a lot of time checking my 1 port setup across 1-300MHz with various test resistances and reactances to prove it can measure impedance reliably. I've used an HP8714B 3GHz VNA with SOL cal kit if anyone is interested in the test gear.
Eg I tested a microwave cap (ATC 22pF 800B series) on my test set/VNA and it agreed very closely with the manufacturers S11 data up to 1GHz. Also I tested various known inductors and resistances.
I also decided on the purest test method that would reduce the potential for error. This was to measure and match the two coils at 100MHz for inductance (122nH) on the VNA WITHOUT the suppressor resistor. I took S1p data for both as a record.
Then I removed them and fitted a 100R shunt resistor to the test fixture and took more S1p data of the resistor on its own.
Then I fitted each inductor in turn carefully across this resistor making sure I didn't alter the inductor turn spacing (to keep the inductance constant and maintain a fair test)
I did this for both NiCr and TCWire. This method removes a couple of sources of error and because the results are so close I needed to reduce errors to a minimum.
This is a MUCH more controlled test than my previous one and the results were as I initially expected. The difference was tiny with NiCr. The change in suppressor Rs at 100MHz was about a 3% increase. I repeated this test procedure lots of times and always got about 2 to 4% change in Rs for NiCr. However, I think it would be wise to allow a few percent on top of this as a precaution despite my attention to detail in reducing errors.
The computed Rp change was from 105.4R down to 101.2R for NiCr at 100MHz.
I have S1p data for both inductors, the 100R resistor and also when combined as a suppressor if anyone wants the data files. I've gone to a lot of trouble to ensure I have got good test data here.
I can post up details of my test setup and calibration checks but this one has gone according to theory. The resistor dominates the Rs figure rather than wire type. In fact wire type comes a poor third to subtle changes in winding inductance (as it should).
So after these carefully controlled VNA tests I think this debate is OVER. I can't see any worthwhile advantage with using NiCr wire in a typical RL suppressor at VHF.
This afternoon I spent a lot of time checking my 1 port setup across 1-300MHz with various test resistances and reactances to prove it can measure impedance reliably. I've used an HP8714B 3GHz VNA with SOL cal kit if anyone is interested in the test gear.
Eg I tested a microwave cap (ATC 22pF 800B series) on my test set/VNA and it agreed very closely with the manufacturers S11 data up to 1GHz. Also I tested various known inductors and resistances.
I also decided on the purest test method that would reduce the potential for error. This was to measure and match the two coils at 100MHz for inductance (122nH) on the VNA WITHOUT the suppressor resistor. I took S1p data for both as a record.
Then I removed them and fitted a 100R shunt resistor to the test fixture and took more S1p data of the resistor on its own.
Then I fitted each inductor in turn carefully across this resistor making sure I didn't alter the inductor turn spacing (to keep the inductance constant and maintain a fair test)
I did this for both NiCr and TCWire. This method removes a couple of sources of error and because the results are so close I needed to reduce errors to a minimum.
This is a MUCH more controlled test than my previous one and the results were as I initially expected. The difference was tiny with NiCr. The change in suppressor Rs at 100MHz was about a 3% increase. I repeated this test procedure lots of times and always got about 2 to 4% change in Rs for NiCr. However, I think it would be wise to allow a few percent on top of this as a precaution despite my attention to detail in reducing errors.
The computed Rp change was from 105.4R down to 101.2R for NiCr at 100MHz.
I have S1p data for both inductors, the 100R resistor and also when combined as a suppressor if anyone wants the data files. I've gone to a lot of trouble to ensure I have got good test data here.
I can post up details of my test setup and calibration checks but this one has gone according to theory. The resistor dominates the Rs figure rather than wire type. In fact wire type comes a poor third to subtle changes in winding inductance (as it should).
So after these carefully controlled VNA tests I think this debate is OVER. I can't see any worthwhile advantage with using NiCr wire in a typical RL suppressor at VHF.
You are not the only one that has concluded that the NiCr suppressor doesn't do much for the VHF parasitic. May I offer a theroy that I have suggested to myself for the last few decades? Rich once said that he had a "special" set of Eimac tubes that appeared to have tremendous gain and with those tubes in ( I think it was a TL-922) the amp would spit and spark and carry on badly. He said it was VHF parasitics and once fitted with his NiCr suppressors the thing tamed down and behaved nicely. What does his suppressors using the NiCr look like at the fundamental frequency? Could it be that his suppressors reduce the gain at the operating frquency thus reducing the gain at the VHF frequency? Perhaps you should run the test and see how the suppressor looks at the operating frequency. One design or Rich's suppressors had a 1 ohm resistor with no coil in series with a RL network in series with another 1 ohm resistor without a coil and another RL network. That deisgn appeared to go by the wayside however. What do you say about that sort of network?
Here's an image of some tabulated data taken from the S1p files. Note that the S1p files have 200 data points from 0-300MHz rather than the few shown below.
You can see the difference in Q at 100MHz between the two wire types when there is no resistor. Eg 165 vs 27.5 so this should prove to any doubters that I was using NiCr wire.
I expected these results because the Rp at 100MHz for the NiCr wire alone is about 2100 ohms and so this plays only a minor role compared to the 100R resistor that is shunting it when it comes to defining Rp. In reality the stray reactances in the layout contribute a degree of muddiness to this simple calculation but I tried to keep these reasonably well controlled. That's also why my data looks more realistic than the N7WS data. i.e. My components are closer to ideal lumped devices so you can see the Rp spread is a lot less with frequency compared to the N7WS data. This helps add confidence to the comparison data
So no surprise that an Rp of 2100 ohms has minimal effect on a 100R resistor. That's why I was expecting a change of less than 5% with a switch to NiCr wire. The VNA measured data when the resistor is added agrees pretty well with this initial rough prediction
You are not the only one that has concluded that the NiCr suppressor doesn't do much for the VHF parasitic. May I offer a theroy that I have suggested to myself for the last few decades? Rich once said that he had a "special" set of Eimac tubes that appeared to have tremendous gain and with those tubes in ( I think it was a TL-922) the amp would spit and spark and carry on badly. He said it was VHF parasitics and once fitted with his NiCr suppressors the thing tamed down and behaved nicely. What does his suppressors using the NiCr look like at the fundamental frequency? Could it be that his suppressors reduce the gain at the operating frquency thus reducing the gain at the VHF frequency? Perhaps you should run the test and see how the suppressor looks at the operating frequency. One design or Rich's suppressors had a 1 ohm resistor with no coil in series with a RL network in series with another 1 ohm resistor without a coil and another RL network. That deisgn appeared to go by the wayside however. What do you say about that sort of network?
I'll have to think about that one tomorrow as I've kind of had my fill of this subject for today and I'm off to bed
Originally Posted by G0HZU
