Super TGSL With H Bridge TX Diff RX Quad Sampling

[Davor]

No, you are seeing it correct. The non-inverting input of an op amp presents high impedance to a transistor collector on it's own, but a low impedance with a large capacitor. It is connected to a positive rail to improve PSSR. It thus settles a referent voltage for the op amp, provided there is a missing 47k connected to a signal ground. As a consequence the inverting input is also kept at that very potential by virtue of feedback, and also presents itself as low impedance to the other collector. In this sense the op amp performs as a current to voltage amplifier (transimpedance). As the same differential current passes through both transistors, using a large capacitor is OK. Benefits of low impedance at collectors is that neither Miller, nor Early can spoil neither frequency response, nor linearity.

[Gwil]

Very interesting. Thanks.

G

[Silver Dollar]

Ok Here's the updated schematic!

[Davor]

Now that I'm reminded of the reasons I designed that preamp the way I did (single op amp) and the way you are employing it here with a second op amp to make a balanced output ... there is a better way of making a preamp that uses a differential pair and a pair of op amps to do it all more elegantly. To be honest, I wanted to make a preamp to improve noise of my IGSL, which uses single ended preamp output and unbalanced switchers. This configuration uses balanced preamp. So why not using the best it can offer?

I'll see to it shortly. Just for a preview, it will be based on a "Superbal" configuration that some sources attribute to Birt.



In this configuration op amp outputs are supplying signal in counterphase, and inputs are seeing equal impedance for both common mode and differential input.

[Davor]

OK, this was quick. Below is a schematic with simple balancing. That large capacitor that Gwil asked about is now gone, and instead there is a feedback network identical to the inverting circuitry. I insist in using a well compensated opamp pair, such as TL072 that is cheap and cheerful in this configuration, and can handle unity gain without stability issues.
Using BC337 or any other transistor with small Rbb should give you noise performance near to 1nV/sqrt(Hz). I've put BC847C just because LTspice models give these transistors Rbb=10ohm and BC337 is given Rbb=60ohm, while in reality BC337 is the one with 10ohm, so go with BC337 and try to match them if possible.
E1 device is here merely for mathematical purpose for presenting a differential output voltage.

BTW, shunting a source of a FET with, say, 100k to signal ground should be beneficial.

[Silver Dollar]

I built the H bridge TX today but there was 2 versions. One had 3U3 caps in series with the both ends of the coil.
I built it without these (as I didn't have any 3U3 nonpolarized). I can't try it until I wind a new Coil.

Davor what do those caps do? I tried a spice simulation with and without and didn't see any difference. Might they
help hide differences in the transistors? Or maybe reclaim some of the energy in the coil? It seemed to draw about
the same current in the sim...

Hmm Can't delete attachments...See second one.

[Davor]

Those capacitors turn a regular H-bridge oscillator into a Clapp configuration which produces higher voltages across the coil. As you already discovered, the downside in these particular frequencies is that capacitors tend to be very large. A sort of solution is to place pull down resistors at each side of a coil (other sides of resistors to ground) that won't load the tank too much to reduce it's Q factor (say 10k each), and use low ESR elko's.

I've been playing with a H-bridge lately and came up to a solution that works in a AB->C class fashion. It draws very little current, and it has superb spectral purity. It also has a current stabilising mechanism that relieves the power supply the oscillator currents, making it a much better companion to the rest of the rig. What it doesn't have is some easy way to set the working conditions using a single component, and I'm about to try something to fix it. Coming very soon

BTW, I've given your FET solution another thought, and it dawned on me that you have a spare gate at 4053 that can do that function better than a FET. 4053 gate can switch between signal and ground, which is better and in this case free - one component less to worry about.

[Silver Dollar]

Yes I had thought about using the extra gate in the 4053 too. I actually had it drawn in but then I
though the gate biasing might be different so removed it until I check into that.

[Davor]

I see,
there are a few mistakes there, but easily fixed. First, the other input should go to the ground, say, via a resistor. Gate should be supplied the same way the other gates are supposed to be, with a resistor going to Vcc, say 100k here and 22k elsewhere. These comparators are toggling between Vee and open circuit, so if you do not put pull-up resistors - nothing happens. Diode should be replaced with a short, and R40 removed.

[Silver Dollar]

Oops I forgot those were 393's and so open collector!

[Davor]

That's more like it. You can remove R66.
I'll not play with the shifters tonight as some other things are on my mind, but most probably you'll have a few more changes before it's done.

[Silver Dollar]

Well here's the new schematic. I took the liberty of adding the compression mods
as I was planning on adding those after running mine stock to see how it works.

I'm just waiting for a coil housing from Bulgaria and a 4053 from Hong Kong then
I can do all the mods. I have the H bridge done and will work on the preamp this
weekend.

I wonder if the compression on the GEB channel could go on the first gain stage?
I played around with the spice file and got a 3V amplitude variation with no compression
on the second amp and a 100k pull down on pin 5 of the 393...

[ashefr10]

hello, why ace you added one 4053 after U106 ?

[paku]

This Modify From Old TGSL Board
or remake New Board For TGSL For This Schematic

[Davor]

hello, why ace you added one 4053 after U106 ?

Such solution improves recovery speed as there is no deep transition into negative voltages. This way targets that are millimetres apart are discerned as separate objects.