inductor wound over a resistor?

[Tom45]

I've encountered them in Tektronix 7A26 plugins. They are used as RLC filters in DC supply lines. There are 4 shown in the schematic: 3.2uH on a 10 ohm carbon composition resistor.

The reason I looked into this is that the tantalum to ground fails short. The coil then cooks the 10 ohm resistor releasing an amazing amount of smoke. Had that happen on more than one 7A26. I bought a bunch of 10 ohm carbon composition resistors and a spool of fine wire and rewound my own. After replacing the failed tantalums, as well as all other tantalums that hadn't failed yet. The voltage rating of the originals was rather low. I think I used 35 volt replacements.

I took the closeup photo so that I could count the turns on an intact coil. 47 for those playing along at home.

[coppercone2]

I have a impedance analyzer now, maybe I will desolder it and measure it for fun to determine the L1 values (that parts list is just too far gone). I just need a fixture. The best I could do right now is put it in a BNC to female banana adapter (screw it under binding posts). Not sure how high up that will go, but I need to fix one of the bands in the impedance analyzer (60-110MHz is broken, waiting on PCB). I measure a mica capacitor OK using the banana jack adapter). I don't think its the worst fixture in the world.. but it should make working on this amp alot more fun in the future, because I see specifically what I am replacing the resistors with (i.e. compare parasitics of new parts to old drifted stuff that is out of spec)

with modern equipment I find usually I can fit a much higher voltage rating capacitor in the same spot, even with tantalums, and still have a size reduction.

[G0HZU]

Be careful if you change the 60uF caps (eg C14 C16) as these are probably solid tant caps and I would expect these caps to have low ESR and low package inductance and they will perform well over a huge RF bandwidth. A modern solid tant might be even better but you would need to prove this before swapping any caps in the RF amp section.

The input emitter follower will have a high input shunt resistance and probably has a few pF input (shunt) capacitance and this Cp will probably be quite constant over a huge bandwidth. So to cancel this (5pF?) capacitance and to obtain a good 50 ohm input match to the amplifier you would ideally need to fit a 50 ohm shunt resistor in parallel with a (negative) shunt capacitor of -5pF  across the input of the amplifier. One way to achieve both objectives is to put a small inductor (L1) in series with a resistor of just over 50 ohms  (R9 59 ohms).

The parallel equivalent of this network is going to be very similar to a -5pF cap in parallel with a resistance of about 50 ohms.

For example, 18nH (L1) in series with 59 ohms (R9) looks similar to -5pF in parallel with a 60 ohms resistor and this holds true across LF and up into VHF.

60 ohms is a bit higher than 50 ohms but note there is also an 1800 ohm resistor in parallel (R10)  and there is also (499+55 = 554 ohms) in parallel.

60 || 1800 || 554 = 52.6 ohms. This is very close to 50 ohms and there will also be some parallel resistance from the BJT. This will have been designed by HP to get really close to 50 ohms input impedance.

Therefore, the value of L1 will be chosen to deliver the correct negative shunt capacitance that offsets the input shunt capacitance of the emitter follower. Note that these capacitances are shunt capacitances Cp in parallel with shunt resistance Rp and not series capacitance Cs in series with resistance Rs.

[coppercone2]

I think those tantalums should be fine no? I thought they basically have no wear/age mechanism and they fail because power supply designers did not limit them well, or carbon comps that limit them degrade to low impedance that do not limit them enough?

Like if C2 was tantalum that might be a concern, but C14? The current levels in this thing seem really low, it does not use a SMPSU.

[G0HZU]

Yes, I'd hope they would be fine if they are solid tants. If they are failing it will be fairly obvious because the response of the amplifier will change.

Do you really want to mess with L1? I'd be tempted to leave it in place and just play with a 56 ohm resistor in series with (say) a 16nH inductor and (for a learning exercise) see if you can prove that this produces  -5pF in parallel with about 56 ohms on your jig/analyser. If it doesn't then maybe your jig could need to be optimised in some way.

I'm pretty sure the amp has been designed this way in order to compensate/offset the input capacitance of the emitter follower and to obtain a good input match for a 50 ohms source. My guess is that the engineer at HP probably experimented with several NPN and PNP emitter follower transistor types before settling on the one used in the final design. The one that is easiest to compensate would probably be the one that was chosen for the HP 461A amplifier.

[coppercone2]

you think it might break from desoldering or something? I guess messing with something that varnished is a concern. The problem I have is that I need to make something like a lever detent switch to press the button on the attenuator if I ever intend to get this amplifier working again, which is a shop project more then anything

(I don't like the mechanism of the attenuator trigger, not one bit, i find it insulting), but yeah I guess I can try to make that circuit myself now that I can measure stuff in VHF with the kind of numbers you are giving me since its just curiosity.

Those long resistors like to crack BTW, there was a cracked one in the attenuator, I guess heating those parts up for no reason is a bad idea..

[G0HZU]

I've never seen inside the HP461A so I don't really know how fragile these parts are. However, see below for a simulation to show that 56 ohms in series with just under 16nH  is quite similar to 57 ohms in parallel with -5pF. The -5pF capacitor turns pink in the simulator if it is a negative capacitor.

You can see why HP would have adopted this compensation approach using a 59 ohm resistor in series with a small inductance. Of course I don't know if the input capacitance of the emitter follower is 5pF. It could be 3pF or it could be 7pF or something different. However, there will be a series RL network that can compensate it reasonably well.

The value of 15.6nH in the simulation below is quite critical. If I change it to 18.6nH it changes the shunt reactance by 20%. I guess that's another reason to leave L1 well alone for now at least.

[coppercone2]

yes, when I tried to make a dead bug oscillator (low UHF) I noticed that those little coils (which I tried to wind around drill bit blanks as a former, and no it did not work) were very sensitive (some low values in the nH range were needed by the schematic, if it is infact a coil that is soldered over a dummy resistor, then I can see how the solder joint connecting the resistor getting desoldered and repositioning the coil could be a problem

https://www.analog.com/en/analog-dialogue/articles/does-the-assembly-orientation-of-an-smps-inductor-affect-emissions.html

better not mess with it


https://article.murata.com/en-global/article/basic-facts-about-inductors-lesson-6


The analysis you did of the circuit to show the sensitivity to component values is a very interesting technique, I guess it should be obvious that its a 20% change, but when units are that small its easy for that to get over looked, I definitely remember like 2-3 nH shifts occurring from basically nothing at all with those tiny coils I wound. Given how much care it looks like HP went to in order to stabilize the input impedance its not worth messing with it

[G0HZU]

Yes, I agree it's probably best to leave L1 alone for now. In case anyone finds this stuff interesting I've rescaled the plot below to show capacitance and it shows how 56R in series with 15.7nH can synthesise a (negative) -5pF shunt capacitance that would both cancel the 5pF of an emitter follower and it would also terminate it quite well with a shunt Rp of about 50 ohms.

I think what you see in the final HP461A circuit is the result of the work of a keen engineer who probably spent quite a bit of time swapping out different emitter follower transistor types and would also have found ways to arrange the compensation network to try and squeeze out the best match to 50 ohms across 1kHz through 150MHz. I suspect it could have been quite rewarding to improve the match because this improves whatever input impedance graph is shown in the specifications for this amplifier. A feather in the cap of the engineer!

So the final solution for R9 and L1 could be wound or arranged in a special way to get the best possible match up at VHF. This might make it a bit strange to look at. I can't comment much more because I've not seen R9 and L1 myself.

[coppercone2]

thinking about my previous UHF circuit attempt, I think that the 10-20 nF inductor would fit that size on the circuit ( i played with the coil designer program alot) for that thickness. I tried to make mine thinner but at that width I bet its exactly that. I wanted to go smaller diameter then what HP has in there for my prototype and I think that I ended up with a turns count I did not like manufacturing.

[David Hess]

The resistor body is a convenient and inexpensive coil form for low values of inductance, and the resistance can be lowered to apply a controlled amount of dampening.

Tektronix did this in the 1960s, 1970s, and 1980s for CLC decoupling.

[coppercone2]

Are these resistors always carbon composition? Once you think about another resistor kind, it seems like you have a coil over a coil and that seems weird. I wanted to do some experiments but the first thing I realized is that all my resistors are some sort of coil structure. That is starting to remind me of a transformer

[David Hess]

Are these resistors always carbon composition? Once you think about another resistor kind, it seems like you have a coil over a coil and that seems weird. I wanted to do some experiments but the first thing I realized is that all my resistors are some sort of coil structure. That is starting to remind me of a transformer

All or almost all of the examples that I have seen use a carbon composition resistor, but the self inductance of a spiral cut film resistor would be insignificant compared to the inductor.

[TimFox]

In most applications of a helix-cut film resistor, the high resistance of the resistive structure should dominate over the inductive reactance.  In fact, I often see that the internal capacitance (turn-to-turn of the helix) is more important than the inductance with axial-lead resistors.  In old-style carbon composition resistors with high ohmic value, the internal capacitance through the talc grains used to increase the bulk resistivity causes the measured real part (resistance) of the impedance to fall at high frequencies (talc dielectric shorting out carbon grains).

[coppercone2]

you know it might also be interesting to wind some inductors over different tubes and see what happens when they are concentric and also physically parallel. I might as well

[David Hess]

The practical aspect is that it is much easier to wind a coil over the perfect cylinder formed by the molded case over a carbon composition element than the "dog bone" shape of a ceramic rod with end caps.

In the example shown below, the 4 big ones are 3.2 microhenries and 10 ohms and feed 10 microfarad solid tantalum capacitors.  Tektronix wound them themselves.

[coppercone2]

Has anyone seen a inductor wound over a capacitor?

[TimFox]

I have seen coils wound around piston trimmer capacitors to make a tunable circuit.
Note that the coil won't be happy if a closed metal cylinder is inside it, which looks like a shorted turn.

[fourfathom]

Has anyone seen a inductor wound over a capacitor?

I've seen "traps" (parallel L/C tuned circuits used to isolate multi-band antenna sections) made from a coil of coax, shorted at one end and the center-conductor open at the other, wound on a plastic pipe form.  It's physically short enough that the distributed transmission-line effects are minimal; the L is the coiled coax shield, and the C is the internal center-conductor-to-shield capacitance. 

Does that count?

[coppercone2]

Has anyone seen a inductor wound over a capacitor?

I've seen "traps" (parallel L/C tuned circuits used to isolate multi-band antenna sections) made from a coil of coax, shorted at one end and the center-conductor open at the other, wound on a plastic pipe form.  It's physically short enough that the distributed transmission-line effects are minimal; the L is the coiled coax shield, and the C is the internal center-conductor-to-shield capacitance. 

Does that count?

i guess, I was wondering what is the likely hood of coming upon something like a foil rolled up capacitor with a coil around it though. Not sure if a tight spiral counts as a shorted turn.

Maybe I will try it and see what happens before and after later tonight, before that magnet wire dies of old age

[coppercone2]

Well I wound like 30 turns of fine wire, maybe more, around a 22nF axial foil capacitor, I think polystyrene (the clear one you can see the spiral through)

When hooked up to the meter it said 20nF at 1Mhz with 0.2 ohms and capacitance started reading negative around 8MHz

When I soldered the coil over it, at 1MHz it said 4nF + 10 ohms, and at 5MHz it read similar to the unmodified capacitor (the ohms dropped back down to 0.3 and capacitance rose), but today is not a day to write things down on a note pad and I don't remember the numbers. But that is like a notch of some kind right, since 4nF and 10 ohms conducts less then 20nF and 0.2 ohms and it looks like the textbook notch filter of LC parallel in series with a load, so by happen stance it looks like I made some kind of low frequency notch filter by winding 30awg kynar wire over a capacitor.

[Karel]

I would like to read the servicemanual of the HP 461A (pdf).
Is there anybody who can post a link or send me a copy?
Thanks.

[Robert763]

A few oldies...

(Attachment Link)

The two on the left and the one on the bottom were recovered by tube b&w TV sets. The two on the top right (sorry one is damaged  ) were commercial units for hobbists/repair, bought in the late seventies. One of them is covered with vax, that was quite common.

Those are not wound on resistors. They are wound on ferrite rod cores that have axial wire leads bonded to them. Completly different component.

Winding on a resistor is as others have said generally used to give a low Q (at self resonance) coil. The resisor has to be low inductance. This is easy with carbon composition but they are hard to find. Modern lossy ferrite "EMC" beads and sleeves have displaced this "trick".

[coppercone2]

just for clarification the schematic of the part says L because I was reading it backwards when I Made the thread... however it does look like someone just wrapped a bunch of wire over a resistor and varnished it.