Precision voltage divider

[kelly0303]

Hello! I need a voltage divider where the input is between -10 and 10 V and the output is between -4 and 4 V. I would need something with a bandwidth of 1 (maybe even 10) MHz. Can someone recommend me a reliable product (the price plays a role, too, so being cheaper would be also useful)? Thank you!

 

[DaveE]

I think what you are describing is two (well chosen) resistors. Can you elaborate on your specs that make this circuit special? We also need to know about the source and load impedance characteristics, particularly capacitance.

 

[berkeman]

How "precise" do you need it? 1%, 0.1%, better? What about temperature compensation? Is this for the laser project?

In terms of bandwidth, you can use the trick used in oscilloscope probes to get good BW even through a voltage divider:

(Cp is the variable capacitance in this case)
http://ecelabs.njit.edu/ece291/lab7.php

 

[Baluncore]

I need a voltage divider where the input is between -10 and 10 V and the output is between -4 and 4 V.

You need to describe your application, and explain what exactly you mean by a "Precision voltage divider".
Will the divider have a fixed ratio, or must that ratio be precisely controlled. Is the 1 MHz bandwidth required of the signal path, and of the variable divider ratio control path?

There are a few alternative possibilities that might meet your specifications.
1. A voltage controlled amplifier.
2. A four quadrant linear analogue multiplier.
3. A multiplying D to A converter, designed for high-speed signal applications.
Each has strengths and weaknesses, so is suited to different applications.

 

[TonyStewart]

Why do you need this, will help us understand the best solution and if there are any other problems. e.g. load, interface, error tolerance (f)

 

[kelly0303]

Hello all, thank you for help and sorry for the lack of details (please let me know if more is needed after this, too). The signal comes from this servo. The impedance seems to be 50 Ohm (same as the laser, which is where the signal goes). The laser frequency changes by 100 MHz/V, and I need it controlled at the 10 kHz level or below so (I think), the noise level should be at the 0.1 mV.

The servo is used for 2 purposes. One is to scan the laser frequency using a triangle wave (for that one I can in principle just reduce by hand the amplitude to +/- 4V). The second one is after I lock the laser (to an external cavity). The servo will send a voltage trying to keep the laser locked to the cavity, and I want to make sure no signal bigger than 4 V is sent (for example if the lock is completely lost, a high voltage might be sent to try to bring the laser back to the lock position). It is this locking phase that I want to limit to +/- 4 V using a voltage divider. Ideally I would like to buy one (e.g. something that accounts/reduces temperature variations) rather than solder it myself. Thank you!

 

[DaveE]

OK... I'm still a bit confused by your specs.
If the servo can only output up to ±10V with a 50Ω source impedance, then when you connect it to a 50Ω load, it will only make ±5V. Then you want a voltage gain of 80%, or a power attenuation of 2dB.

If it really can make ±10V into a 50Ω load, then you want a voltage gain of 40%, or a power attenuation of 8dB.

You can buy attenuators from any good supplier (digikey, for example). You get to choose things like PCB mount or connector type, etc.

Here are a couple that might work (depending on your real requirements of course):
https://www.digikey.com/en/products/detail/mini-circuits/HAT-8A/19186516
https://www.digikey.com/en/products/detail/mini-circuits/HAT-2A/19186424

PS: Oops, maybe not. These guys can only handle 1W. See @Baluncore's post below.

 

[TonyStewart]

There are several input & output specs :

Max Input Voltage DC Level	±500	mV
Max Input Voltage Signal Amplitude	±500	mV
Ramp Amplitude (Max)	±5	V
Output Voltage (main and Aux)	±10	V
RF Output Max Amplitude (-PL only) ±75 mV RF Output Impedance (-PL only) 50 Ω

Error Input (BNC)

datasheet inputs:
This is the input for the error signal. In SIDE LOCK mode, the signal is amplified by 26 dB and summed with the DC OFFSET and DC OFFSET INPUT (back panel). In PEAK LOCK mode, the ERROR INPUT is demodulated by the dither frequency and is then amplified by 26 dB and summed with the DC OFFSET and DC OFFSET INPUT (back panel). In both modes, the amplified signal can be seen with the DC ERROR MONITOR and the AC ERROR MONITOR.
Gain Sign (two-position switch)
The lock switch has three positions. The lowest is the RAMP, which connects the internal ramp to the SERVO OUTPUT causing the laser to sweep. The amplitude of the sweep is controlled with RAMP AMP knob. In the center position (UNLOCK) the ramp is disconnected and zero volts is output to SERVO OUTPUT. In the top position (LOCK) the loop filter is engaged.

========================================


I assume you have -PL option. Please confirm your requirements and list below all physical and electrical functions. A block diagram would be nice.

Did you mean you have no one who can solder?
"reduces temperature variations) rather than solder it myself."

 

[kelly0303]

There are several input & output specs :

Max Input Voltage DC Level	±500	mV
Max Input Voltage Signal Amplitude	±500	mV
Ramp Amplitude (Max)	±5	V
Output Voltage (main and Aux)	±10	V
RF Output Max Amplitude (-PL only) ±75 mV RF Output Impedance (-PL only) 50 Ω

View attachment 326926
Error Input (BNC)

datasheet inputs:
This is the input for the error signal. In SIDE LOCK mode, the signal is amplified by 26 dB and summed with the DC OFFSET and DC OFFSET INPUT (back panel). In PEAK LOCK mode, the ERROR INPUT is demodulated by the dither frequency and is then amplified by 26 dB and summed with the DC OFFSET and DC OFFSET INPUT (back panel). In both modes, the amplified signal can be seen with the DC ERROR MONITOR and the AC ERROR MONITOR.
Gain Sign (two-position switch)
The lock switch has three positions. The lowest is the RAMP, which connects the internal ramp to the SERVO OUTPUT causing the laser to sweep. The amplitude of the sweep is controlled with RAMP AMP knob. In the center position (UNLOCK) the ramp is disconnected and zero volts is output to SERVO OUTPUT. In the top position (LOCK) the loop filter is engaged.
View attachment 326927
========================================


I assume you have -PL option. Please confirm your requirements and list below all physical and electrical functions. A block diagram would be nice.

Did you mean you have no one who can solder?
"reduces temperature variations) rather than solder it myself."

The output that goes to my laser is the Servo Out and the maximum voltage coming out of that is +/- 10 V. So I need to reduce that maximum to 4 V. I can solder, but I am not sure if I can solder it in such a way that I can not be sensitive to temperature variation for example. Actually I am not sure how a precision voltage divider is different than just 2 soldered resistors, but I assume it's not that simple, no?

 

[DaveE]

The output that goes to my laser is the Servo Out and the maximum voltage coming out of that is +/- 10 V. So I need to reduce that maximum to 4 V. I can solder, but I am not sure if I can solder it in such a way that I can not be sensitive to temperature variation for example.

Why exactly are you so worried about "precision" and "temperature variation" for this attenuator? It's inside of a feedback loop, right? Locking doesn't require precision, you're scanning over a wide range. Once it's locked won't the servo fix any drift or calibration issues? Please explain what you mean by "precision".

Note that the generic resistors (i.e. the cheap ones, about $0.01 each in volume), we used in our designs everywhere, typically were ±1% and 100ppm/^oC. That's normally pretty good except for unusual situations. Granted, we didn't make cheap stuff; you couldn't buy it on amazon.com.

I am not sure how a precision voltage divider is different than just 2 soldered resistors, but I assume it's not that simple, no?

It really could be that simple. The ones you buy from places like mini-circuits are designed for real RF, so they will be 3 resistors in a π configuration (or maybe Bridged Tee) to avoid impedance mismatches and the consequences. But I'm not convinced you need to care too much about that at ≤10MHz.

 

[Baluncore]

Just thinking on the back of an envelope.

For an RF coaxial 50 ohm attenuator;
10 volts / 50 ohms = 200 mA input.
6 volts; 1.2 watts in the attenuator.
4 volts; 0.8 watts in the load.

Maybe a simple series resistor of 75 ohms could drop the extra 6 volts.
10 volts / (75 + 50 ) ohms = 80 mA.
4 volts; 0.46 watt in the 75 ohm resistor.

 

[berkeman]

Maybe a simple series resistor of 75 ohms could drop the extra 6 volts.
10 volts / (75 + 50 ) ohms = 80 mA.
4 volts; 0.46 watt in the 75 ohm resistor.

So just use a 1 Watt, 75 Ohm resistor as a first try? Nice idea.

(actually, maybe use a 5 Watt resistor to keep it from getting too hot to the touch)

 

[TonyStewart]

You have been given good advice. it is not hard to make a small series R attenuator for 10 MHz Bandwidth can tolerate a few cm of inductance from the lead lengthSince 10V is the maximum with 0.46W we would expect a 0.5W to rise 100'C and a 1W resistor to rise 50'C which is OK for the resistor but might burn a finger and a 5W resistor would only rise 10'C which is cool to the touch.

But then the amount of time your error signal is 10Vdc is going to be small and this reduces to some small value when it is locked. So the choice of power rated series 75 Ohm resistor which will attenuate fine 50/(50+75)=40% of input. Anything from 0.5W to 5W will work fine.
Perhaps you can salvage a short BNC cable and splice the series resistor in series. The microwave worthy parts are overpriced because of the radius ratio tolerances which define the impedance errors and return loss.How you create this is a simple mechanical problem. To solder the resistor between BNC female jacks or how perform this simple task is up to your imagination. A BNC splitter or attenuator might be convenient but $$. You can order attenuators from Pasternack, to avoid this. But that is $65 overkill for a << $1 resistor and connectors. I would just splice a cable and insert the 75 Ohm 0.5W to 1W resistor and coat with epoxy or PU for strength.

 

[DaveE]

But then the amount of time your error signal is 10Vdc is going to be small and this reduces to some small value when it is locked.

Umm... are you sure? I thought it was the freq. input to the laser; what we, in the controls world, would call the plant input, after the error amp and compensation. I think you have to plan on the loop locking near and staying near 10V. IDK, maybe the rest of the system parameters (which we don't know) make this unlikely.

 

[TonyStewart]

Umm... are you sure? I thought it was the freq. input to the laser; what we, in the controls world, would call the plant input, after the error amp and compensation. I think you have to plan on the loop locking near and staying near 10V. IDK, maybe the rest of the system parameters (which we don't know) make this unlikely.

You're right, I'm not sure of the block diagram for this application. But we design for worst case anyways.

snip from datasheet
"
The SERVO OUTPUT is the output from the loop filter when in LOCK mode, zero volts when in UNLOCK mode, and a DC balanced triangle wave when in RAMP mode. The ramp can be shifted with respect to 0V using the ramp centering functionality.Auxiliary Servo OutputThe AUXILIARY SERVO OUTPUT is generated from integrating the SERVO OUTPUT. Its purpose is to supply a correction signal to drive the SERVO OUTPUT to zero. When used with Vescent DBR lasers and the D2-105 Laser Controller, the AUXILIARY SERVO OUTPUT can be connected to the TEMP SERVO IN to adjust the laser diode temperature to keep the feedback laser current constant. Similarly, AUXILIARY SERVO OUTPUT can drive a PZT on an external-cavity laser diode to keep the laser diode current constant"

 

[TonyStewart]

Kelly would leave less uncertainty and guessing if a proper block diagram and all pertinent specs were provided in the initial question.

Please confirm the following uncertainties.
1. Is the laser input = 50 Ohms ?
2. If +/-10V out is with 50 ohm load then unloaded is +/-20V
If 1&2. are true then you need a 150 ohm series metal film 2W resistor to a 50 Ohm load.
If 2 is false, then you need a 25 ohm series resistor 2W to the 50 ohm load laser.
If 1&2 are false then you need 2 resistors.