Transformer-coupled dpot

FF67D67D34458B072894D837 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit

So I'm doing a spin of my swoopy nanoamp photoreceiver board, to make it smaller and more versatile.

The original QL01 has two pots, one for offset and one for HF boost. The HF boost is a lead-lag network that attenuates the voltage the FB resistor sees, reducing the effect of its parallel capacitance.

The first one can easily be a dpot, but the second one is a bit more of a challenge. It's not too hard to level-shift unidirectional I2c up to the op amp's output

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs
Loading thread data ...

I guess the object of the level shifter is to interface with the floating dpot. The sim can't find the TLV431 model, which I guess is just a sort of floating power supply.

A transformer is an interesting way to isolate a variable part, like a pot or a varicap or something.

We use

COILCRAFT PWB1010LB

which is a 780 uH 1:1 transformer spec'd for 3 KHz to 125 MHz.

but we use it for signal coupling, not "impedance isolation."

Why not a varicap to ground, inside the feedback loop? Sorta in place of C3.

--
John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 
 Click to see the full signature
Reply to
jlarkin

You might use a digital capacitor to tweak HF response.

formatting link

--
John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 
 Click to see the full signature
Reply to
jlarkin

Those are very pretty. I sort of need at least a 10V swing, though.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Slide it a little closer to the summing point maybe?

--
John Larkin         Highland Technology, Inc 

Science teaches us to doubt. 
 Click to see the full signature
Reply to
jlarkin

Hard to do without destroying the SNR, because the larger Johnson noise of the small segment of the feedback resistor can flow through the lead-lag network. The same thing happens if you try splitting the resistor and bootstrapping the midpoint via a lower-resistance string--the SNR is the same as if you just used the first resistor by itself.

If transformer coupling a dpot doesn't work, I'll probably wind up switching capacitors to ground with a DG409 or something like that. The caps would be in the hundreds of pF, in series with 120 ohms or so, which wouldn't tax the HV mux's capabilities too much.

I'd really like to use a dpot though, because that can be programmed at test time whereas using a mux would mean I'd need a micro on a hitherto all-analogue board.

Mini-Circuits sells transformers that work down to 4 kHz at 50 ohms, so ISTM they should work at 40 kHz at 500 ohms. We'll see.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Seems to me that the transformer doesn't do much. Couldn't you use the dpot as a pot, with one end grounded, one end driven from the amp (with a resistor maybe) and the wiper driving the compensation?

Reply to
John Larkin

Those are very pretty. I sort of need at least a 10V swing, though.

From a compensation POV, that's a feature rather than a bug. ;)

The transformer more than halves the voltage applied to the dpot, so I can use AD5273BRJZ1's with a 6V supply and hang one end on the +3.3V rail. It also eliminates the need to level-shift the I2C.

I'd have to drop the power supplies way down to avoid frying the dpot, so the feedback resistor would have to go way down as well, which would hurt the SNR at low currents. The Johnson noise of the feedback resistor equals the shot noise when it's dropping 2kT/e (50 mV at room temperature). With 10M, that's 5 nA, but with 2M it's 25 nA, and I care about that quite a lot.

I have some of these on order. Both have 10 kHz minimum frequency quoted.

which is 1:1:1, , and which is 1:4

Using my trusty AADE, I measured a T622 1:1:1 (100 kHz-200 MHz). It was

35 uH per winding open, 35 nH with one winding shorted. Since

L_sc/L_oc = 1-k**2,

that works out to k = 0.9995.

If the ratios are the same, the 10 kHz 4:1 will be about 350 uH on the primary and 1.4 mH on the secondary. That leads to a bit of tilt on the pulse tops unless I halve the impedance level of the lead-lag network. I can probably use a fixed resistor there to do most of the work, and run the dpot with the two ends shorted, which gives vernier action near half-scale.

It might be easier to fix the tilt in the second stage by giving a bit of a LF gain cut.

Fun.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

If you're going to program it at test, could you put two or three DPOTs in series (so less voltage across each one), and power them either with photovoltaic isolators, or some fancy op-amp bootstrapping circuit. Yes the SPI level shifters could get interesting, but those could be a one-off monstrosity residing in your tester, behind the pogo pins or whatever.

Reply to
Chris Jones

Not a bad idea--I'm not at all above such things. In a tester, it could be three Bus Pirates hung off USB isolators, so nobody would even have to know. ;)

This is one of those things that comes up a fair amount though, so it would be useful to have a solution that would work in a box with a uC as well.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Oh, there's another trick: a tiny surface-mount dip switch and 4 caps makes a 4-bit capacitive DAC.

Or a tiny screwdriver-adjusted variable capacitor.

Both are cheap and low noise and require no software.

formatting link

We pay 26 cents for that one.

Reply to
John Larkin

I'm not at all above that sort of thing either. I could even use a hex dip switch to mimic a variable cap. ;)

Agreed. Part of what we're doing is to make our gizmos attractive for people to license, and as you know there's a lot of prejudice out there against manual trims. That's not completely unreasonable when the person doing the adjusting has no idea how the circuit works--it's easy to encapsulate the know-how in the test software.

For protos and proof-of-concept systems I often use gimmick caps to adjust amplifier peaking.

Long obsolete, unfortunately.

Our existing trimpot works okay but doesn't play that well with ATE. (You used to be able to get GPIB-controlled screwdrivers for that sort of thing.)

Plus, needing to hang an AC tweak on some out-of-the-way circuit node is something that comes up a fair amount round here.

The transformers will be here today, so maybe next week I'll give it a whirl. We're ordering boards probably tomorrow, so it'll have to be trimpots for this batch.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Maybe that's why we have a ton of them. Probably an EOL buy.

I still think you could use the dpot as a grounded pot, but I don't entirely understand your situation.

Reply to
John Larkin

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.