RF signal amplification

[whiz115]

What type of device are L2 and L3? Do you have a web site where I can find a data sheet?

i'm posting the data on my last post, also a photo.

 

[mneary]

Everybody knows what VK200 is, but nobody has a proper data sheet. I guess I'll have to assume it's 550 ohms resistive.

I haven't been able to find the base characteristics of the 2N3866.

 

[whiz115]

is BF199 biased ok? mneary said something about it...i gave him an answer but i didn't got any response.

how can i fix the bias for the U2?

is it really needed R3 to be bypassed with a capacitor?

the output PI filter as i previously said is wrong since i want it for 75ohm load and it's not calculated that well, i also want the impdedance match circuit to be with fixed values, since the RF amplifier is going to be used around 100- 108MHz...l'l try to correct them as soon as i will understand the java applet marcbarker gave me with the smith charts etc.

unless someone can speed up the process for me...

 

[marcbarker]

I don't get why you say that. 2:1 VSWR is like a 25% reduction in power if I recall correctly.

25% drop is only worth a few dB's. With, say an overall path loss something like 80 dB for example, having few dB difference doesn't figure much at all. There's more variation going on due to multipath reception. In a decent radio link, the AGC will accomodate variations in RX signal strength.

<..> you can't use a dummy load that only works at one frequency and has to be adjusted to make it work at all.

Isn't this exactly the identical same behaviour you will get from almost any Antenna?

<..> wirewound resistors are massive value inductors at 100MHz.

With series-cap, it's still a series-tuned circuit, and ends up as resistive impedance. (@ the freq of interest)


If you're only ever working with one frequency, it probably helps too, because it's a 'high-level wavemeter', if there was ever a gross harmonic emission problem, it would readily show up as an increase in VSWR (being out of band).

Meanwhile, in Amateur Radio as we all know, the Dummy Load is an investment, so any dummy load is expected to be broadband! Of course, given the choice, a non-inductive load is preferred, a ready-made one for 2m is ideal.

wirewound resistors at 100 Mhz?

Admittedly, getting a home-made dummy load to work at 600 MHz (Analogue TV transmitter) was tricky. I used a single light bulb & ceramic ('behive') trimmer in series.

 

[marcbarker]

Whiz? Are you still working on this circuit (in the meantime)?

I think I can see what you're trying to do. The series resistors were a good idea, they prevent the transistors blowing up too easily. I take it U1 is amplifying OK-ish?

I think an empirical approach helps. This is what I do with this kind of circuit.

U2 is just not biased right at all, it'll probably run hot too. You need to have a at least a low value resistor between U2's E & B. Start with 47 Ohm.

Temporarily link out L3 and disconnect L4 from U2 C. The resistor R3 wants to be this 'broad-band dummy load' the conversation was about earlier. You definately can't use a bulb here (not unless working with a lower frequency).

Move trimmer C4 to across U2's e-b resistor. It's your choice whether you want to use 'L', or 'Pi' matching. Pi is more versatile. There needs to be a little bit more work here. In meantime for inspiration look at the example schematic that was posted earlier.

And roughly tune C3, C4 for maximum supply current (being dissipated in the 22ohm). This is the start of the empirical stage, the experience you get here is golden. You'll find the routine here almost identical to those Agilent Java 'tuning' applets I posted.

You might might find that one of both trimmers run out of travel, you'll have to suss out that one yourself. You're allowed to 'adjust' L1. It doesn't have to be 180.00 nH and it won't work any better either being exactly that value. But the adjustment of L1C3C4 is critical.

Remove link from L3, reconnect L4. Now you work further along the circuit, doing the same routine with the adjustments.

 

[unclejed613]

Everybody knows what VK200 is, but nobody has a proper data sheet. I guess I'll have to assume it's 550 ohms resistive.

I haven't been able to find the base characteristics of the 2N3866.

the VK200 is actually just the ferrite core with 6 holes in it. what the actual inductance is depends on how many turns you put through it. i've even seen (with very small wire) 75 to 300 ohm baluns wound on these little cores,

 

[unclejed613]

here's a more complete data sheet with the input (base) reflection chart

 

[mneary]

Thanks for the data, been looking all over for it. Until now, I claimed I could find anything. S-parameters are only given down to 150, OP is designing for 100. I guess that's DC on this data sheet.

If I read the graphs correctly, our base should be in the neighborhood of 25-j8 ohms?

I thought I saw something like the VK200 used as baluns, but when I saw the impedance data points it occurred to me that they seemed kind of lossy for that kind of service.

 

[unclejed613]

that's probably why they're used as chokes. rf chokes always seem to be chosen for being a bit lossy.

as far as the data sheet goes, Datasheet Archive - Free Datasheet Search Engine - PDF Datasheets - Data Sheet - Datasheet - Application Note - Free Texas Instruments Samples been using it for years, and rarely fail to find what i'm looking for in one form or another. once you get used to identifying the source material, and have an idea what you're looking for, you can find almost anything there, except for vacuum tube data sheets.

 

[mneary]

I have found half a dozen data sheets for the 2N3866, the miracle of your discovery is the Smith chart and AC characteristics. All the pretty sheets from Philips, NXP and the rest would like us to use it for DC and something else for RF.

 

[marcbarker]

Motorola App Notes are useful. Part of the good old "AN-267" app note is reprinted in the appendix of this one a hold-your-hand RF amplifer design.
https://www.electro-tech-online.com/custompdfs/2009/08/an791.pdf
This example they choose a Q of 8 for the matching network. They're using a MRF247 with a z-in of 2.06 +j 1.08 (parallel) @ 2m

I find it much quicker to find the values of matching network empirically rather than do the sums and try get it right first time. I'll only do rough sums just to get me into the ball-park values first. After a while you kind of 'know' what the values would be, without having to do the sums.

 

[whiz115]

U2 is just not biased right at all, it'll probably run hot too. You need to have a at least a low value resistor between U2's E & B. Start with 47 Ohm.

you mean a 47ohm at the emitter?

Temporarily link out L3 and disconnect L4 from U2 C. The resistor R3 wants to be this 'broad-band dummy load' the conversation was about earlier. You definately can't use a bulb here (not unless working with a lower frequency).

what does "link out" mean? i don't know that phrasal verb.. english isn't my native language..

R3 as well as the rest of the resistors are going to be a 1% tolerance...i don't know if their material is going to be suitable for low inductance...

Move trimmer C4 to across U2's e-b resistor. It's your choice whether you want to use 'L', or 'Pi' matching. Pi is more versatile. There needs to be a little bit more work here. In meantime for inspiration look at the example schematic that was posted earlier.

i'm getting confused where do you want me to move the C4?

the java applet from besserNet is very interesting and useful but i don't know about smith charts as well as there are many parameters which i don't know what i should do so i can calculate a correct PI filter. can you tell me more about it so i can understand what's going on?!

the example from agilent is very fun.... put the ball in the hole...then what?

if you want read my reply to mmary at the previous page and if you like answer them since mmary didn't...

the VK200 is actually just the ferrite core with 6 holes in it. what the actual inductance is depends on how many turns you put through it. i've even seen (with very small wire) 75 to 300 ohm baluns wound on these little cores,

indeed it's a ferrite with 6 holes...but what makes you think it is sold without the wire?


the most important thing is if these chokes are suitable were i intend to place them on my circuit.

 

[unclejed613]

"link out" means "put a short circuit across it"

the 47 ohm resistor goes from emitter to base.

1% tolerance resistors aren't neccesary, 5% tolerance parts are usually used. use metal film or carbon film resistors. wirewounds are (as mentioned previously) inductive.

"indeed it's a ferrite with 6 holes...but what makes you think it is sold without the wire?"
companies like Ferroxcube make the ferrite cores and sell them. that's why every time the VK200 was mentioned in data sheets and app notes, the wire gauge and number of turns was also described.

use the chokes for isolating the RF from DC sources. that's why it doesn't matter if they are lossy. all it has to do is act as a high impedance to RF while maintaining a very low resistance to DC, and that includes going from ground to the base of U2. U2's base is at DC ground potential, but the full signal level of RF from the previous stage is being fed into the base of U2 (well maybe a little bit less, since it's going through an impedance matcher)

 

[Hero999]

1% tolerance resistors aren't neccesary, 5% tolerance parts are usually used. use metal film or carbon film resistors.

I agree 5% is fine, obviously there's nothing wrong with using 1% if it's all you've got to hand but it is a bit overkill.

companies like Ferroxcube make the ferrite cores and sell them. that's why every time the VK200 was mentioned in data sheets and app notes, the wire gauge and number of turns was also described.

That's right, you wind your own choke.

 

[whiz115]

unclejed613 thanks for the explaining...

an update of schematic.. marcbarker i'm not sure i did exactly as you
told me...check the design and tell me..

 

[marcbarker]

I've attached your schematic which I have 'marked up' (another phrasal verb!)
I didn't realise that english is not your native tongue. What is your native tongue?

I've changed the matching network into a 'pi' matching network. What you ought to do is use roughly calculated values here. Otherwise you often find that a tuning capacitor is not enough pF's.

Any stray capacitance in the circuit and inductance such as lead inductance (and any unavoidable self inductance) will be present in the circuit. You normally are supposed to select components so that these effects are not significant to cause "parasitic effects" (also known as 'sprogging' in slang). You would never be able to eliminate all of these effects, even if you used the best components you can find. What you can do is allow the adjustment of the Matching Network to accommodate this stray capacitance and inductance.

Carbon film resistors (e.g. 1% types) are somewhat inductive. It's because there is a spiral etched into the carbon film. The old carbon composition ones (e.g. 10%) are blocks of carbon.

Ferrite is a wonderful material. It's so 'forgiving' too, if you make design mistakes. When making an RF choke, you'll get the broadest 'flattest' band (lowest Q is good) if you have a single ferrite bead with a wire inside it, but sometimes it is not enough impedance. To increase the impedance, you can get the multiple hole ferrite. That is like connecting ferrite beads in series. I think it's something like 30 Ohms per turn or something (in the datasheet). Too much impedance is bad.

Typical use for ferrite is to put a DC short circuit between the base and emitter of a transistor (operating in 'class C'). The RF doesn't 'see' this short circuit.

The java applets gives you a very accurate taste of what it is like for you when you are tuning a matching network.

 

[whiz115]

I didn't realise that english is not your native tongue. What is your native tongue?

is this a compliment? i'm greek...

I've changed the matching network into a 'pi' matching network. What you ought to do is use roughly calculated values here. Otherwise you often find that a tuning capacitor is not enough pF's.

i want to get rid the variable capacitors so it's a bless if they aren't needed that much.

The old carbon composition ones (e.g. 10%) are blocks of carbon.

you mean the usual 10% resistors are fine?


marcbarker i have difficulties understanding this weird biasing with the U2...
do you think you can help with that? or i'm asking too much?


by the way i have started the designing of the PCB...

 

[Hero999]

Carbon film resistors (e.g. 1% types) are somewhat inductive. It's because there is a spiral etched into the carbon film. The old carbon composition ones (e.g. 10%) are blocks of carbon.

I don't think I've seen 1% carbon film, most carbon film is 5%.

Yes they are slightly inductive, I did think about mentioning it, but then I realised that the inductance must only be about 10nH which makes no difference even at VHF unless the resistance is very low.

You can still buy carbon composition resistors but they're expensive. I would recommend thin film resistors, most are SMT but the 1206 packages are fairly easy to solder.

EDIT:

you mean the usual 10% resistors are fine?

Yes 10% will probably be fine but I think you've missed his point about the type of resistors used. 10% resistors aren't that common nowadays.

Carbon composition reistors look like this:
**broken link removed**

Carbon film resistors look like this:
**broken link removed**

 

[whiz115]

Carbon composition reistors look like this

i know them but i don't have them... probably i can find some!

 

[marcbarker]

I don't think I've seen 1% carbon film, most carbon film is 5%.

I think I meant to of said: "metal film" under the paint, it has a spiral etched into it

Any idea what are those holsworthy/meggit" ones? or the "mrs25" ?