Turns ratio of RF transformers.

[Wendymott]

Hi Peeps. Now please understand..when it comes to deep RF Theory and maths.. I am a complete novice.. and I wondered if you learned peeps might give me a helping hand.
I have gone back to my multiband HF SSB Transceiver.. and I would like to optomise or nearly, using... my component stock, without excessive purchasing for "one off" bits.
80M Filter. I wound T1 with 35T on a 421 Toko former and it is resonant at 3.65 Mhz with a 56pf cap. The reactance I calculated was 780R. Thus if I want a 50R input winding..I did the following.. 780R / 50R = 15.6 Thus 35T / 15.6 = 2.25T
Obviously I would use 2T as 2.25T would be difficult.. However... is my reasoning correct please ??
Also... another problem... The T2 secondary will feed a Plessey SL610 RF Amp, but I cannot decide what the turns ratio should be as there is no definite input impedance of the SL610.. The data sheet states input Admittance of approx "0" at 3 Mhz... but I am unsure how I treat this.. is the 50R suitable.
Now I know I am using antique devices.. but it is better to "recycle" than buy new...and it is not intended as a super douper transceiver as the noise levels at my QTH are abysmal ...about S7 on 40M
Thanks in antisipation

[G8HQP Dave]

80M Filter. I wound T1 with 35T on a 421 Toko former and it is resonant at 3.65 Mhz with a 56pf cap. The reactance I calculated was 780R. Thus if I want a 50R input winding..I did the following.. 780R / 50R = 15.6 Thus 35T / 15.6 = 2.25T

Two problems with that calculation: it gets the impedance ratio wrong; it ignores Q.

Impedance varies as the square of the turns ratio, so if you want to turn 780R to 50R with a transformer then you need sqrt(15.6)=3.95 turns ratio.

However, you probably don't have an impedance of 780R to start with. That is just the reactance of the coil and capacitor. The circuit impedance depends on Q and whatever is loading the tuned circuit. As things stand, if you applied an external load of 50R to the circuit via the correctly calculated ratio for 780R the net result would be a Q around 1 so very little selectivity.

I don't think you can avoid gaining the necessary understanding before calculating what loaded Q is needed and how to match to that. There is a balance to be struck between selectivity and loss, so there is not a 'right' answer which someone else can give you.

[Wendymott]

Hi Dave. Many thanks for the reply.... I guessed it would not be as straight forward as I stated..in fact the primary turns finished up at 5T......I will post photo's from the spectrum analyser in tracking mode.. when I have rebuilt the 20/15 and 10 M filters... they were in a pre construction mode ....a couple of days should suffice.

I am asking why have transformers in a filter, surly a standard 50ohm in 50ohm out (or whatever impedance your design has) design will do?

[G0HZU_JMR]

Also... another problem... The T2 secondary will feed a Plessey SL610 RF Amp, but I cannot decide what the turns ratio should be as there is no definite input impedance of the SL610.. The data sheet states input Admittance of approx "0" at 3 Mhz... but I am unsure how I treat this.. is the 50R suitable.

The datasheet gives an admittance chart in mmho (with the output o/c) and I've shown a doctored version below for just the SL610 or SL1610 device.

I've also shown a crude equivalent model that only applies across about 3-20MHz. This is 2500 ohm in parallel with 4.3pF. This seems to match the datasheet. My chart is in mho and not mmho so the 1 mmho line on the datasheet corresponds with my 1e-3 mho line.

But you can see the model does fairly well across 3-20MHz. However, the real part of the admittance turns negative up above 25MHz and the chip will be prone to instability up here.

is the 50R suitable.

Probably... I think these devices are meant to be driven by a source of just under 1k ohm for best noise figure. But this optimal source impedance will only give about a 3dB NF. With a 50R source driving it the NF will be worse, but it should still be about 6dB NF. So no big deal on the HF band.
Note that this doesn't mean you can fit a 50R shunt (filter) termination at the input of the chip and get a 6dB NF. The NF will be much worse if you do this. I think the source can be 50R but you get the 6dB NF if the 50R source is only terminated by the high Z SL610.

I never really got on with the SL61x series of chips from Plessey but this was in my early years at work. However, I'd still be nervous to use one today because the device is so prone to instability.

[G0HZU_JMR]

The other thing with the SL61x series is they are voltage amplifiers and the gain specs reflect this. The SL610 has a voltage gain of 10 and the device has a fairly high input impedance and a low (slightly negative) output impedance.

In a typical application a series resistor of nominally 56R is fitted at the output to help negate the negative resistance here and if you power matched the input I think the SL610 will have a lot of power gain. It could be close to 30dB. i.e. if you had an L match or a transformer to match from 50R to the high Z input and had the 56R at the output I think you can get a lot of power gain. Far too much for an RF amplifier in a typical HF receiver.

[Wendymott]

Hi Merlinmaxwell. I dont really understand your answer.. Both Tuned elements are a bandpass filter which when parallel tuned will be high impedance.. My question was.. the turns ratio per filter to allow 50 ohms input impedance of T1/T3/T5 etc to minimise loading of the filter... and the output turns ratio to match the SL610.
Hi Jeremy. I do understand that the SL6xx series are now a lot "old hat". and were designed in the 60's for the Clansman series of radio's...but I have them..so they will be used...may not be as efficient design wise as other devices... but will suit my purpose. Thanks for your comments though and will be applied, especially for the 610. In due course I will post my schematic and the filter characteristics. I do understand the 40 metre filter should be better than this design, and you kindly did lots of calculations using torroids, but.... in this case I wanted "simple" and the 9 Mhz point is 20 db down.

[G0HZU_JMR]

OK, if you want to use the SL610C I can offer some more bits of info but this is from my distant memory and a quick refresh of the datasheet. So the following may contain nuts

Because this is a voltage amplifier with a high Z input and voltage source output the gain and noise figure performance will vary a lot depending on how you interface to it. The same applies to stability. I do think it is a good idea to fit a series 56R resistor at the output. This will make the (slightly negative) output impedance look like 50R and it will help a bit with stability.

Have a look at the two circuits below. The one with the shunt 56R at the input will terminate the filter nicely and the gain of the overall stage will be about 14dB (I think). However, the noise figure is likely to be 12-13dB. Not good! One bonus is I'd expect it to be very stable like this. Probably unconditionally stable. However, if you removed the input 56R shunt I think the gain will go up by 6dB and the noise figure will improve to maybe 7dB.

But and it's a huge BUT... it will be nowhere near unconditionally stable and when connected to switches and BPFs at the input it might prove a challenge to keep it all stable.

The other circuit shows what happens if you power match to the input of the SL610C with an L match or a transformer. The noise figure will be lower and it might be 6-7dB but the power gain of the overall stage will be up towards 29dB. A huge difference. However, it will probably be stable with the L match at the input, but I'd expect this to be marginal. With a transformer at the input I'd expect it to be prone to instability depending on the transformer design and what is connected before the transformer.

[G0HZU_JMR]

If you are wondering why there is such a difference in gain between the two amplifiers then I think I can explain below:

Because the SL610C is a voltage amplifier it will amplify the voltage at its input pin by 10.

So a simple model for the first amplifier would be unity voltage gain (wrt an ideal 50R termination) into the 56R load followed by a gain of ten in the SL610C followed by a division of two in the series 56R to 50R port at the output.

So the voltage gain would be 1 x 10 x 0.5 = 5

Because the amplifier is 50R in and out the power gain = 20*log(5) = 14dB gain in the overall amplifier.

A simple model for the second amplifier would be a voltage gain of 7 in the L match (to 2500 ohm) followed by a gain of ten in the SL610C followed by a division of two in the series 56R to 50R port at the output.

So the voltage gain would be 7 x 10 x 0.5 = 35

Because the amplifier is 50R in and out the power gain = 20*log(35) = 30.88dB gain in the overall amplifier. However, I'd expect it to be closer to 29dB in reality. But that really is just a guess.

My advice would be to try and find a compromise design between the two extremes above. Maybe match to a 200R-500R termination resistor at the SL610C input for example. The noise figure would be better than the first amplifier and the gain wouldn't be as scarily high as the second amplifier? A good compromise?

Note that if you try and use the SL610C up at 21MHz or 28MHz it will be much harder to keep it stable if you connect filters or L matches or transformers directly at its input. It might need a fixed attenuator after the SL610C to keep it unconditionally stable if you want to play with transformers/filters connected directly at the input with no input termination resistor. The datasheet usually shows no termination resistor at the input and this is to help keep the noise figure low. But it does make it harder to maintain unconditional stability like this.

[Wendymott]

Hi Jeremy... Thank you for your time and trouble... I will have a look at your suggestions in the next couple of days...I want to finish the L/C filters first... they are a separate module... it is strange that the Plessey data book and its application notes seem to take the very simple ways of using these beasties.... and that is what I was relying on.. at one point I did try using X3 SL612's in a series mode after the 9 Mhz filter, but it seemed way too wild for me..with the AGC generator in circuit... maybe mixing technologies such as the SL series and SMD devices is not as good as leaded components..may have extra parasitic components which calm things down. Dunno.

[G8HQP Dave]

You can't finish the filters until you have decided how you want to terminate them. The load on the second tuned circuit will partly determine what impedance you are matching the first tuned circuit to, and the amount of coupling needed between them. You need to decide the loaded Q (anything less than the unloaded Q!), which decides the bandwidth and loss. You need to decide the coupling - probably around or just below critical coupling. These are design decisions which we cannot make for you.

[G0HZU_JMR]

Yes, it's worth designing the filter and transformer and SL610C interface all in one go and it's up to Wendy to decide how selective the filters are and what impedance she wants to present to the SL610C.

If it helps, I had a rummage through an ancient logbook at work today. We used the SL1611C in an AGC IF amp about 30 years ago and there was some info photocopied from a Plessey design note about these amps.

If you look at the circuit below, this is what Plessey refer to as the 'normal' amplifier setup and it is supposed to be able to prevent oscillation even with tuned input circuits up at 28MHz for example. With the 47R in series at the output and the 1000R shunt resistor it appears the amplifier will be unconditionally stable across its full range. But this was for the SL1611C. However I suspect it also applies to the SL(1)610C as well.

So the circuit below should be quite flexible and remain stable. You could experiment with home made filter transformers at the input and vary the tap point to find the best compromise in terms of match and noise figure and gain. If you power match to the 1000R resistor I think the amplifier gain will be up around 24dB or so. I think the noise figure would be about 9dB.

The reason the noise figure is as bad as 12-13dB in the first 14dB gain amplifier I showed you with the 56R input termination is because the datasheet implies the noise figure will be about 6dB when the SL610C is driven from a 50R source. But when the 56R termination is added the source seen by the SL610C is closer to 25R and there is a 6dB loss of signal (but not noise) once the source is terminated because there is a 50R to 56R potential divider. So the noise figure degrades to about 6+6dB = 12dB.

[G0HZU_JMR]

I had a look to see if I could find any SL610C devices at work today but I don't think we ever used this device. But the plastic DIL 8 SL1611C is on our database and I found a short tube of them in the old rotary carousel of parts in customer support. They have a 1985 date code! They have probably been there longer than I have worked at the company.

[Wendymott]

Hi Jeremy... Again many thanks for your input... I attach photo's of the filter block with a SL610 test jig attached... configured as your note above. I also attach the resultant filter characteristics ...
Marker notes.... 80 Metre... marker 1 = 3.5 mhz. marker 3 = 3.79 Mhz and marker 2 = 4.08 mhz
40 Metre. M1 = 7.0 mhz... M2 = 7.21 Mhz.. M3 = 9 Mhz
20 Metre M1 = 14.0 Mhz.. M2 = 14.36 Mhz
Omitted 15 metre as there is only space for 5 pictures
10 Metre.. M1 = 28.04 mhz. M2 = 30.06 Mhz.. M3 = 31.22 Mhz

Yes I know the 9 Mhz point is not as attenuated as it might be in an ideal world..but its ok for now.

David.. Please remember.. this is not intended as a theoretical project... more a "fly by the pants" project.. where I can get as good as I can without the maths, or minimal maths as it is not my subject. But thanks anyway.