Telecnatron Forum

A variable gain hybrid-cascode amplifier stage followed by separate feedback amplifier stage, as used for amplifying the output from AD9850 DDS module as the basis for an HF signal generator.

Hello.

I found this article to be very interesting and quite useful given the increasing popularity of these low cost DDS modules. My comment is regarding the feedback amplifier you reference in Experimental Methods in RF Design. The circuit model for this amplifier is Figure 2.57 and Figure 2.64 which shows the negative feedback bypass capacitor in parallel to the bias resistor, both in series with the feedback resistor R-f to the transistor base. I noticed your arrangement was quite different. Was this by design?

Best regards,
ChasW

I can not get any usable gain. I built with good RF practices. I used 3/8" - #43 torrid. Output Z is more than 50 ohms. Primary winding is about 50 uH. I even have problems simulating with given values in LTSpice. I have spent a couple days now, checking my Spice and Breadboard? Too bad, I spent a lot of time? Mike

First of all thank you for this contribution. Secondly, I wanted to if I could use an output impedence of 8Ohms instead of 470Ohms in the amplifier stage. The thing is I have a Hammond audio transformer lying around with impedance ratio of 8:600 and my final load is of 600. I was wondering if I could use it.

Interesting article however can you confirm the part number of the Schottky diode (2N5819 in schematics) as when I lookup that part it appears to be a PNP-BJT. Also, can you confirm the size of wire you used to wind the transformer (12:4 turns ratio)?

Hello Cameron,

Yes, the Schottky diode is incorrectly shown as a 2N5819 BJT. Thanks for pointing that out. It's just for reverse polarity protection, so pretty much any diode with current rating over 200ma or so should do, Use as Schottky for lower forward voltage drop, or omit it altogether if you are confident that you wont reverse positive and negative.

Regarding the wire gauge on the 12:4 transformer: IIRC I used thin wire-wrap wire on a 43-type material binocular core, which is probably not ideal as the wire has a thick plastic coating which could be more lossy than enamel coated. I don't think that it's too critical though: as thick as will fit perhaps? Anywhere from 20AWG to 26AWG should be just fine.

Cheers,
Steve VK2AAV

Hello Steve,

I'm not very familiar with this topic, but I need an amplifier for the AD9850. First, the frequency is constant around 27 MHz. Is it possible to get a higher gain at this frequency if the other frequencies are not used?

Can I use 100n ceramic in SMD 0805 for the capacitors?

The next question is how to wind the transformer. I use the FR 37-43 Core and did it like this:

[photos.app.goo.gl]

but when I was done I wondered about the start of the winding ?! I hope you understand what i mean. It can be wrapped on the right and left.

The inside diameter of the core is approx. 4.7 mm and the outside diameter is 9.5 mm, the height is 3.2 mm.

I would be very happy to receive an answer and clarify the questions. Best regards from Germany to Australia.

Peter

Hello
My Hycas Buffer did not give significant gain,
until I suscpect and replace the source resistor with 500 Ohm trimpot.
Start to work when the source resistor around 97 Ohm or smaller.
Maybe due to FET pichoff and Idss characteristic.


Yours,
Gendhengiradio

Hello, I built this as VGA for a DDS VFO/sweeper, but could not find the J310 FETs, so I tried many other types I have in my parts box, BF245, BF256, 2N4117, 2N5457, PF5102, etc. And I tried hand selecting the highest IDSS, low VP, high VP etc, no voltage gain. At most near unity, like -3 to -5 dB and worse.
Then I tried 12:4 12:3 12:1 transformer ratios, no gain. Then I tried more windings 30:10 and lower, no voltage gain. I tried a few different cores. It worked well as an attenuator with losses. The FET also introduces a sharp second harmonic, but not much higher order ones.

So, after simulating lots of variations of R, different coupling coefficients, different coils, I decided to ditch the transformer and replace it with a 470 ohm resistor, followed by an emitter follower. Now, this works and has a very wide bandwidth and above all; more gain. And still a very low output impedance, it drives a 50 ohm load up to 30 MHz up to ~15 dBm. This is with jelly bean BC547 transistors and a mediocre BF245. This increases the current needed by this amplifier by a lot!
So, if you are stuck with building this VGA, try adding an emitter follower (or two). Do not forget a series output resistor on the emitter of a few ohms to reduce parasitic oscillations when a capacitive load is attached.

Then I tried getting the set point of the FET at it's ideal point with multiple diodes, this is working to reduce second and higgher harmonics! If you get this working, try biasing the FET up or down for lowest distortion. This means you'll have to redo the gain setting of the potentiometer as the range shifts.
This circuit really could benefit from an anti-log pot, but those can be mimicked by a linear pot with a parallel resistor of about 1/3 the value of the pot. This means the relative position of the pot can correspond more to an absolute dB scale.

I measured quite a nice flat frequency response up to 28 MHz, with a slow rolloff of about 1 dB per 3 MHz. Above 28 it rolls of very steeply. Also, with large input and output caps it can go as low 1 Hz... That is quite insane. To power this combination, you'll need to filter the power supply a lot more than with the transformer.

Well, those are my two cents.

All in all, your post was a great inspiration, thank you so much!

Ben