7
\$\begingroup\$

I'm using Microchip TC1071 in a circuit to provide an adjustable output voltage (appears to be pin compatible with LT1761-SD). My plan is to use a digital potentiometer (Microchip MCP4251) to control the output voltage within over some range. The schematics and layout are on solder pad here. The adjustable regulator circuits are in the upper left quadrant of both the schematics and layout.

I've got it all hooked up per their typical application circuit and it doesn't seem to work the way I'd expected it to. I have an 0805 100nF ceramic capacitor near the input voltage (pin 1) of the regulator. I put this 10uF electrolytic capacitor near the output voltage (pin 5). The high side of the voltage divider is connected to my digital potentiometer. The low side is connected to a 56 kOhm 0805 resistor. The \$\mathrm{\overline{SHDN}}\$ (pin 3) measures 5V as does the input voltage.

So I blew the digital potentiometer off the board and hooked up a fixed 47kOhm resistor. If I read the datasheet correctly, the voltage I should get out of the regulator in this configuration is {1.2 * [( 47 / 56 ) + 1]} = 2.21V. Instead, I measure about 1.6V. I'm staring at this datasheet and am somewhat baffled. Are my capacitors inappropriate to support this regulator (seems like there are some ESR requirements, but I can't tell what the ESR of my caps is)? They suggest using somewhat larger resistors for the voltage divider:

The ohmic values of these resistors should be between 470K and 3M to minimize bleeder current.

But could this actually impact the accuracy of the output voltage (that much)? Does anyone have any experience with these adjustable regulators that can offer me some practical advice on getting the thing to work, and what I might be doing wrong?

Edit

I should note that I only need to source 50mA max, though 100mA would give some more breathing room.

Edit

Once I resolved the netlist error with some trace cuts and softwires, I was able to control the voltage as I had intended per the design.

\$\endgroup\$
  • \$\begingroup\$ Should I just try and switch to this (btw cheaper) part? digikey.com/product-detail/en/LM317LD13TR/497-1572-1-ND/592073 \$\endgroup\$ – vicatcu Jul 22 '12 at 3:21
  • \$\begingroup\$ Can you check the output of the regulator with an oscilloscope to make sure it isn't oscillating? \$\endgroup\$ – W5VO Jul 22 '12 at 4:35
  • \$\begingroup\$ @W5VO yes I can check the output with a scope, but not until monday... \$\endgroup\$ – vicatcu Jul 22 '12 at 4:39
  • \$\begingroup\$ The LM317 is adjustable, but isn't an LDO like the Microchip part. You'll have less output range. \$\endgroup\$ – Adam Lawrence Jul 22 '12 at 4:44
  • 1
    \$\begingroup\$ I get an error page trying to view your capacitor link \$\endgroup\$ – W5VO Jul 22 '12 at 14:46
6
\$\begingroup\$

It looks to me like a netlist/layout error:

schematic screenshot

If you are placing your fixed resistor on the dig pot footprint, it would also be messed up. Looks like you accidentally wired the CO+ and NO2- together and the NO2+ and CO- together so the LDO(Vout-Vadj) is not across a resistance as you believe.

\$\endgroup\$
  • \$\begingroup\$ Wow, that is an amazing catch. Thank, I'll have to get the exacto knife and soft wires out and see if uncrossing these wires gets it running properly. Will report back. \$\endgroup\$ – vicatcu Jul 22 '12 at 21:49
  • \$\begingroup\$ :) have had a few similar mistakes in my past \$\endgroup\$ – justing Jul 22 '12 at 22:11
4
\$\begingroup\$

The only reason to go with a high-resistance divider is to ensure that you're not wasting power unnecessarily - the LDO most certainly shouldn't care how stiff the divider is.

I don't have a clear understanding why the programmable pot 'blew off the board'. Do you? Did you connect the power connections incorrectly?

Inappropriate ESR would cause the regulator to oscillate, which sometimes causes a DC reading to look incorrect if you're measuring it with a multimeter. It shouldn't cause a DC setpoint error. Do you have the ability to scope the rail?

Other than that, other reasons why the output is lower than expected is loss of headroom (shouldn't be an issue with 5V in) or overload (exceeding the specified output current).

\$\endgroup\$
  • \$\begingroup\$ it didn't blow off the board, I hot-air blew it off to remove it as a variable... \$\endgroup\$ – vicatcu Jul 22 '12 at 4:58
  • \$\begingroup\$ yes i will scope the rail next chance I get (probably monday) \$\endgroup\$ – vicatcu Jul 22 '12 at 4:59
  • \$\begingroup\$ I linked the capacitor manufacturer specs, do you see an ESR in there anywhere? I didn't... \$\endgroup\$ – vicatcu Jul 22 '12 at 5:33
  • \$\begingroup\$ It actually is, in a manner of speaking. The loss tangent is the ratio of ESR to capacitive reactance, and is usually expressed in percent. 0.2% = ESR / |Xc| (at 120Hz) so the ESR should be around 265 milliohms. \$\endgroup\$ – Adam Lawrence Jul 23 '12 at 21:16
4
\$\begingroup\$

Do you have a chance to check the output with an oscilloscope? It is possible that your LDO runs unstable. The reason might have to do with your output capacitor's ESR. Please look at page 6 of the datasheet. This LDO (like many others) is only stable when the output capacitor's ESR is in a certain range:

LDO - ESR Tunnel of Death

Some call this region between the lower and upper limits for the ESR the LDO's tunnel of death. While adding a low-ESR ceramic capacitor is often a good idea, it may be bad here.

Related: Choosing a Tantalum based on Min. ESR - LDO Stability

Also: One thing I would worry about with a digital pot between Vout and Ref of an LDO is the digipot's default value after power up, before it is told what to do. It might happen that you get an undesired, and, depending on the load, potentially harmful output voltage before the whole thing gets into proper regulation.

\$\endgroup\$
  • \$\begingroup\$ I sized the low side resistor so that when the Digit-pot powers up to mid-scale it sets the LDO to about the right voltage... \$\endgroup\$ – vicatcu Jul 22 '12 at 18:31
2
\$\begingroup\$

If you do not need LDO the LM317 is certainly reliable. Yes the LDO is also low current making it very high impedance so bias resistors of 47KΩ are no good. I do not understand why you did not choose within the range of 470KΩ and 3MΩ as directed. Yes it is that sensitive. LM317 is not so fussy since it is not that high Ω or low drop out.. but 2.5V drop or so from Vin to max Vout. Ok here. they use 2.50 Vref.

BTW the OEM cap specs ought to indicate ESR. If too low it cannot remain stable since voltage feedback loop gain is delayed due to low R ratio. Too high and you get excess ripple. You can measure with any 50Ω sig gen... Use 1 Ω series R and measure voltage drop where voltage is equal between R and cap and then compute impedance at that freq. many other methods...using fixed F & measure R ratio of 1 Ω and Cap ESR. Choose series resistor near expected ESR for best sensitivity.

added

  • YOu need to consider range of resistance for Digital Pot

  • Vreg input bias current. eg. LM317 is Adjustment Pin Current 50 typ 100 max [μA]

  • V drop In-Out ~2V Worst case @-50C @ 300mA typ.

  • Min Load current may apply.. Spec says 3.5 typ 5 max @ 40V drop.. so 2V drop Min load =?

enter image description here


There is now a wide variety of LM317's so keep in mind they are not all the same and not all suppliers are identical crossovers unless spec'd

I used NSC which is now owned by TI and so may I suggest if you still want to use this approach, consider what load you need and choose accordingly.
http://www.ti.com/lit/ds/symlink/lm317l-n.pdf ( low power )

http://www.ti.com/sitesearch/docs/universalsearch.tsp?searchTerm=lm317&linkId=1&x=0&y=0 ( universal selector for LM317's)

You may prefer to simply use PWM or a simple DAC with current buffer, the latter which I suggested in alt answer.. Or just define why you need a digital supply, there may be even simpler solutions.

\$\endgroup\$
  • \$\begingroup\$ the reason I didn't pick a 470kOhm to 3MOhm resistor is because the largest digipot I could find was 100kOhm.... \$\endgroup\$ – vicatcu Jul 22 '12 at 4:35
  • \$\begingroup\$ Then use the LM317 \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jul 22 '12 at 5:06
  • 2
    \$\begingroup\$ He says he uses 5 V in. The LM317 is specified for a minimum input-output difference of 3 V. Minimum out is 1.25 V. So the LM317 will give vicatcu only a 0.75 V range. Not good. \$\endgroup\$ – stevenvh Jul 22 '12 at 10:44
  • \$\begingroup\$ THat is incorrect and might only come close to a 3V drop at 2A if that were even possible. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jul 22 '12 at 10:52
  • 3
    \$\begingroup\$ I'm looking at the TI datasheet. (If you have another one you should add a link in your answer.) Recommended operating conditions: 3 V difference minimum. Line regulation and reference voltage are specified for this 3 V difference. That means that you cannot rely on the reference voltage being between 1.20 V and 1.30 V if you don't respect that. Then anything is possible. \$\endgroup\$ – stevenvh Jul 22 '12 at 11:04
1
\$\begingroup\$

I think you should scrap the idea of a Digital pot and 3 terminal adjustable regulator and simply use a simple 4/8/12/16 bit DAC and a current buffer unity gain.

Choose a serial or parallel DAC as you prefer with resolution required and define your load and V range with power source options, if you need a buffer selection assistance.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.