Large Value Inductor Question

[PsychMan]

Hi Folks,

I have been given a circuit for a single transistor RF amp that can be bolted onto an am modulator to increase its output (responsibly of course). The modulator in question uses a CD4053 as the final, and the circuit is designed to go between the CD4053's output and the onboard filter network for antenna matching.

The circuit calls for a 22mH(!) collector choke. The problem I have is despite assembling the circuit as described, performance is poor (worse than with no amplifier circuit), and the 22mH choke is being cooked to death. Looking into the specs of a lot of such value inductors, often state a SRF of 200khz - not a lot of use for medium wave. I'm assuming this is why the amplifier is working very poorly. Or am I missing something?

Ill dig the circuit out and add it here if anyone is interested.

Cheers
Adam

[Skywave]

You state that "the 22mH choke is being cooked to death".

I believe that the heat being produced to do that can only be due to the resistance of the choke and / or excessive current flowing through it. (Power = I²*R). For the former cause, a physically larger choke - with thicker wire - but maintaining the required inductance is required. For the latter case, the device sinking that current through that choke is excessive, which leads us to a different fault, not a fault due to the choke alone.
A look at the circuit will be instructive.

It will also be interesting to see what others here think.

Al.

[PsychMan]

Sure Al, I wondered about that too, but I cant seem to find much choice when it comes to such high value inductors. Anything reasonably big with bigger wire either has low SRF values or lower inductance. So I wonder if these values are in fact correct. Attached is a sketch of the circuit, excuse the poor drawing!

My understanding of the collector chokes main role is to ensure the RF goes out to the ATU and not into other parts of the circuit like the power supply.

[dave_n_t]

Originally Posted by PsychMan

Sure Al, I wondered about that too, but I cant seem to find much choice when it comes to such high value inductors. Anything reasonably big with bigger wire either has low SRF values or lower inductance. So I wonder if these values are in fact correct. Attached is a sketch of the circuit, excuse the poor drawing!

My understanding of the collector chokes main role is to ensure the RF goes out to the ATU and not into other parts of the circuit like the power supply.

I tend to agree with Al (Skywave). Could you measure the (DC) voltage at the emitter; and tell us the choke's (DC) resistance? That give an idea of how much power the choke is dissipating purely due to the DC component.


dave

[Aub]

Could it be 2.2mH ?

[PsychMan]

I will take some voltage readings, sure.

I was told it was 22mH, but I do wonder actually.... There doesn't seem an off the shelf choke that exists with the characteristics required. I'd imagine it would be pretty huge if it did.

My understanding of its role (aside from a slight effect on collector current)- is to stop RF getting where its not wanted, and would also require it to be significantly higher in inductance than seen in the antenna matching network, is that a fair assumption?

2.2mH being much higher than anything needed to match a short antenna wire..

[kalee20]

22mH chokes can indeed be made with SRF above 200kHz.

SRF itself is not an issue, it's the actual impedance which is.

However, in the current circuit, I'd suggest 2.2mH would be adequate. If the ATU can tune it to resonance, fine!

Scope photos of waveforms would be a great help.

As the present choke is being cooked to death, what's its DC resistance, and what current is flowing?

[Skywave]

That circuit diagram: now it may be me over-looking something fundamental or perhaps my arithmetic is awry, (both are quite likely ) but to me, that cct. looks all wrong . . . . I'll explain . . .

The 1N914 clamps the 'top end' of the 1 mH choke to 0.7 v., since the 1N914 is a silicon diode and is fwd. biassed.
Now the voltage at the base of the 2N3904 to 0v., i.e. (Vb-0v.) will equal the Vbe of that transistor plus Ve, where Ve is the emitter voltage to 0v.

Therefore:
(Vb-0v.) = Vbe + Ve,
so:
(Vb-0v) - 0.7 = Ve.
For Ve > 0v., (which must be so if the transistor is conducting),
(Vb-0v) - 0.7 > 0.
Hence, (Vb-0v) > 0.7 v.
However, the 'top end' of the 1 mH choke is at 0.7 v.,
so: (Vb-0v) cannot be > 0.7 v.


A simpler way to look at all that is this:
Neglect the base current of the 2N3904.
The 1N914 makes (Vb-0v) = 0.7 v.
But Vbe of the 2N3904 = 0.7 v. (if it's conducting - which, presumably, is the whole idea!)
Hence voltage across the emitter R = 0v.
Hence, the 2N3904 is not conducting.
Hence, Q.: where is all the current - that is flowing through the 22mH - choke going?

Nothing makes any sense here - or so it seems to me.

Anyone here care to enlighten my darkness?

Al.

[Julesomega]

Class C?

[trsomian]

Do not forget that a choke used at a frequency greater than it self resonant frequency behaves as a capacitor, at resonance it is neither capacitive or inductive. You do not state the core material of the choke, could the over heating be due to eddy current losses??

[Radio Wrangler]

Skywave's posted the relevant question.

The bias arrangement in that circuit is awful. It's a sort often used to bias class B or AB RF power amplifiers, but in those cases there is great care over thermal bonding of the diode and the transistor. The idea of voltage drop versus diode current is a bit of a simplification, seen aat the scale of the silicon, it's a matter of voltage drop versus current density. If you put two similar diodes in parallel and run them to the same voltage, then each ought to be passing the same current, so you get double the current of one diode and together the pair have junctions of double the area of one diode.

Things are a bit more complicated in the base junction of the transistor, but it is possible to set the current in the 1N914 to bias the transistor into conducting some amount of quiescent current. THat ten Ohm emitter resistor is an attempt to curb its instabilities.

The question I'd add is that for a pantry transmitter is a circuit intended to wring the last bit of power out of a device really necessary? Low power amps can be done easily enough with full class A and bias as stable as you could want.

The 22mH inductor was probably an old sectional wound RF choke from someone's junk box. Not a component often seen in the catalogues nowadays.

David

[PsychMan]

I lashed this up again last night quickly to report back - the choke read 42 ohms, and the voltage across is was 10v . Collector voltage under this scenario was 1.9v and 1v seen at the base.

I had a 2.2mH choke which was physically bigger and read 1.4 ohms. With this in circuit instead, there way less than 1v across it, 12.2v at the collector and 0.4v at base. Will need to experiment further with the biasing and matching as initial tests with some ATUs I've knocked up didn't yield very much output, they were just quick tests however. But it did appear pretty clean audio wise, so isn't affecting the modulation.

So It does look like I was given the wrong value for it and it must be 2.2mH. The person who gave me the circuit I believe used off the shelf components, and as such a 22mH choke doesn't seem to exist, I guess that's it

[PsychMan]

With regards SRF, Kalee and others. My understanding of SRF is that you want the inductors maximum SRF to be considerably HIGHER than the frequencies it will see, is that correct to say? As above the max SRF it will not behave as an inductor?

Id imagine you COULD make a 22mH choke with high enough SRF , and much of it would depend on the type of core material used, while the dc resistance would relate to the gauge of wire used and the diameter of it when wound? The datasheet for this piddly little inductor doesn't state what material is used, as far as I can see. I guess it is made for other applications than this anyway

[PsychMan]

Originally Posted by Radio Wrangler

Skywave's posted the relevant question.

The bias arrangement in that circuit is awful. It's a sort often used to bias class B or AB RF power amplifiers, but in those cases there is great care over thermal bonding of the diode and the transistor. The idea of voltage drop versus diode current is a bit of a simplification, seen aat the scale of the silicon, it's a matter of voltage drop versus current density. If you put two similar diodes in parallel and run them to the same voltage, then each ought to be passing the same current, so you get double the current of one diode and together the pair have junctions of double the area of one diode.

Things are a bit more complicated in the base junction of the transistor, but it is possible to set the current in the 1N914 to bias the transistor into conducting some amount of quiescent current. THat ten Ohm emitter resistor is an attempt to curb its instabilities.

The question I'd add is that for a pantry transmitter is a circuit intended to wring the last bit of power out of a device really necessary? Low power amps can be done easily enough with full class A and bias as stable as you could want.

The 22mH inductor was probably an old sectional wound RF choke from someone's junk box. Not a component often seen in the catalogues nowadays.

David

Thanks David, that's very interesting and I will have to read that a few times and try to understand it better

I only used this circuit as it was suggested as a reasonable Class C improvement to the little transmitter Ive been mucking about with. Id gladly bin it if I could find something that would work better. I don't think I could design one myself at this stage, so just something to muck about and learn with.

This is the device in question if anyone is interested:

https://www.nfor.nl/radioforumservic...PLL_AM_Osc.pdf

[Julesomega]

I'd expect similar values for the base and collector chokes so 2.2mH sounds sensible. 1mH is about 3.5k ohms at 330kHz which is more than enough for circuit impedances around 50 ohms. In fact 2.2mH is if anything a bit high for 1.6MH and is at risk of resonance

[Julesomega]

Are your measured voltages with rf applied? They make no sense for pure DC
Measure again through a resistor between 1-10k shunted to gnd by 10-100nF

[G6Tanuki]

What's the "ATU" circuit you are using?

If it's a classic 'pi'-section then one useful trick is to move the collector-feed choke from the collector-end ot the pi-coil to the antenna-side of the pi-coil, so the collector-current actually flows through the pi-coil.

The advantage here being that the choke is essentially 'decoupled' from the collector by the low-pass effect of the pi-circuit - making spurious oscillations due to self-resonance in the collector- and base-feed chokes/networks a lot less likely.

[This approach also works in valve PAs, and was used in quite a lot of WWII US military gear. It took some time for the idea to be embraced by the ham-community]

[PsychMan]

Ah yes, that was with RF applied. I can disconnect it and take some measurements.

I do have a multimeter probe circuit I use to measure antenna voltages for matching, but had overlooked the fact there is RF in the circuit itself which would mess up the readings

[PsychMan]

Originally Posted by G6Tanuki

What's the "ATU" circuit you are using?

If it's a classic 'pi'-section then one useful trick is to move the collector-feed choke from the collector-end ot the pi-coil to the antenna-side of the pi-coil, so the collector-current actually flows through the pi-coil.

The advantage here being that the choke is essentially 'decoupled' from the collector by the low-pass effect of the pi-circuit - making spurious oscillations due to self-resonance in the collector- and base-feed chokes/networks a lot less likely.

[This approach also works in valve PAs, and was used in quite a lot of WWII US military gear. It took some time for the idea to be embraced by the ham-community]

Interesting! I might have a play with that.

I have a few things I usually use when experimenting like this:

A tapped loading coil wound on large diameter plastic pipe, used in conjunction with a variable capacitor in an L network, OR a little piece of board with a PI network on it. It's quite adjustable with various small inductors with shorting switches across them, switchable capacitance across the input and a variable capacitor on the output. Then measure the antenna voltage using a probe circuit.

With this circuit I had better results with the loading coil rather than the PI network.

I made the coil some time ago and initially never had much luck with it. Most advice I found stated to leave unused turns open, but I found it works very well by shorting the unused turns, or infact grounding them. Not sure why but there must some interaction there when they are left open.

[Skywave]

Ref. my post #8.
I had not realised that this cct. is designed to run other than Class-A. In the light of that, my previous analysis is somewhat amiss. However, having said that, even for Class C, that bias arrangement is truly awful. But why choose Class-C for the required application? Surely Class-A would be quite adequate, even if it meant using a transistor capable of handling more mA and watts than a 2N3904. And as a side-line bonus, the design arithmetic - and consequently the match between predicted behaviour and resultant performance - becomes a lot easier, too!

Al.