LM317 Higher Voltage 90v fixed low current

[marce002]

Hello

I have some LM317 lying around and wanted to make as simplest as posib. a circuit that outputs 90V fixed , low current needed say max 250ma, when the input is around 100v max ... if the input is under 90v well no problem going slightly down .... what would be the simplest circuit approach?



thanks in advance

[Benta]

Bad idea.

[jimmc]

This old National Semi document might help, Linear Brief 47 High Voltage Adjustable Power Supplies
https://www.ti.com/lit/an/snoa648/snoa648.pdf
Jim

[Benta]

This old National Semi document might help, Linear Brief 47 High Voltage Adjustable Power Supplies

What a crazy circuit. Thanks for posting it.

[marce002]

thanks Jimm, very ittle current 25ma at 160V, maybe it is higher at 100v? also VIn is >170v and i only have 100v max

[TimFox]

Here is one of many implementations of the "Maida" HV regulator using a 317.
Google "Maida regulator" for others.

[Konkedout]

Many years ago I designed a production low noise 48V output switching power supply using the LM317.  Main issue is that it is necessary to limit maximum voltage ACROSS any two pins of the device.  My design worked really well.

I had maybe 8 1N4007 diodes in series from input to output, and then (spanning a few of them) I connected the (feedback optocoupler LED in series with a resistor.)  The switcher approximately regulated the voltage across the LM317 and the LM317 regulated the output.  For low noise I had a film capacitor (maybe 470 nF) from ADJ pin to ground.   To protect the IC against sudden output shorts I also had a 1M4007 connected from ADJ pin to Vout.  I don't have the schematic available to show.  This might not work so easily if you are not building a switcher of some type.

I had tried using a TVS instead of the series 1N4007 diodes, but in testing with sudden output shorts, that did not stand up as well as the multiple 1N4007 diodes.

[BrokenYugo]

LR8 and a transistor.

[Zero999]

Bad idea.

Why? The LM317 is a floating regulator. The input voltage is 100V and output 90V, a difference of 10V. It should be fine, so long as it doesn't current limit. If there's a significant capacitance on the output, put a zener or TVS diode between the input and output, to protect it against the brief turn on spike.

[marce002]

Bad idea.

Why? The LM317 is a floating regulator. The input voltage is 100V and output 90V, a difference of 10V. It should be fine, so long as it doesn't current limit. If there's a significant capacitance on the output, put a zener or TVS diode between the input and output, to protect it against the brief turn on spike.

Thanks Zero!! do you have any circuit image of a circuit that does the trick? input 90-110V and output always 90V and in case input < 90 then output=input ..just that

[Kleinstein]

The LB47 circuit is not that bad. For a little more current it just needs an adjustment to the 100 ohm resistor before the regulator and maybe a bit more sturdy power transistor. Here only 100 V to start with make it a bit easier.

[Zero999]

Here's how I would do it.

If there are no capacitors on the output, D1 can be omitted, otherwise it's important, to prevent the voltage rating of the LM317 from being exceeded.

Note, the output voltage will rise, if the input voltage exceeds the output by the breakdown voltage of D1.

[marce002]

Thanks Zero999 but LM317 only can be used up to 37Vol so this 100V input is totally wrong, you meant Tl783 ?! i
I will build and try it !! and comment back...
If I knew any circuit simulator  i would do this test before... (wonder what would happen if):

input < 90V then ---> should be output = input
input > 100 then ---> should be output = 90V

[ejeffrey]

Also if D1 is asked to carry significant current, it will die extremely quickly.  For charging up a small capacitor it's probably fine but if for some reason it ends up supplying the full load current (say someone shorts the output) it will have a brief and explosive life, followed immediately by the lm317.

[Zero999]

Also if D1 is asked to carry significant current, it will die extremely quickly.  For charging up a small capacitor it's probably fine but if for some reason it ends up supplying the full load current (say someone shorts the output) it will have a brief and explosive life, followed immediately by the lm317.

True. It has no short circuit protection. I did warn the original poster about it.
https://www.eevblog.com/forum/beginners/lm317-higher-voltage-90v-fixed-low-current/msg5341190/#msg5341190

Thanks Zero999 genious! i
I will build and try it !! and comment back...
If I knew any circuit simulator  i would do this test before... (wonder what would happen if):

input < 90V then ---> should be output = input
input > 100 then ---> should be output = 90V

The .asc file attached to my previous post, can be simulated in LTSpice, but there's no point because models for voltage regulators are often approximate.

Actually, the output will really be 89.2V, if the resistors are exactly 130R and 9k1, the reference is exactly 1.25V and the current flowing out of the adjust pin is exactly 50µA, but there's no point in aiming for resistor values will give exactly 90V, because the tolerance of the LM317's is too poor. In other words, 89.2V is within 1% tolerance of 90V, whilst the LM317 has a tolerance of 4%.

The LM317 has a certain drop-out voltage which is required to maintain regulation and in this case D1 bypasses it, when the input-output difference exceeds the breakdown voltage, which is 33V in this case.

Assuming the current is 250mA and room temperature and ignoring the 4% tolerance i.e. circuit will give exactly 90V out:

Input < 91.7V: the output will be about 1.7V less than the input.

Input is between 91.7V and 93V, the output will have poor regulation, but should still be close to 90V

Input between 93V & <123V: the output will be 90V and well-regulated.

Input >123V: the output = input - 33V.

[marce002]

thanks very much very complete and clear!!!

[enut11]

This old National Semi document might help, Linear Brief 47 High Voltage Adjustable Power Supplies
https://www.ti.com/lit/an/snoa648/snoa648.pdf
Jim

I built this circuit (Figure 2) and it works well. Just pay attention to the output string (R6+R7) as it dissipates some heat so good quality resistors are required for long term stability.
enut11

https://www.eevblog.com/forum/metrology/afforable-100vdc-reference-for-met-experiments/msg3247074/#msg3247074

[marce002]

Hi everyone

I have taken the proyect into real life, but something is not as expected, i did not use any zener nor caps,  the output is almost identical to the input less 2 vol , for eg if input is 100v, the output is 98,3V (instead close to 90v),  it is like it follows the input -2v , not fixed at 90v at all.. what im doing wrong, if not my resistors values:?

PD do not have exact resistors so i did

1) 4 parallel 470ohm for the 130R
2) 2 series 4.7kohm for the 9K1
 

[Kleinstein]

The resistors values should set the voltage to some 100 V. So maybe it still works. For a test maybe remove one of the 470 ohm resistors or replace with 1 K.

[marce002]

thanks i will try, but the doubt here if the lm317 is limited to 37 vo, why did he allow me to use it in the circuit i have made....

[Zero999]

The LM317 is a floating regulator. All voltages are relative. The voltage rating of 37V, relates difference between the input and output, not the total voltage.

Look at the voltages on the different terminals of the LM317. With 100V in and 90V out, the maximum voltage seen by the LM317 is 100 - 88.75 = 11.25V.


You didn't include the zener diode, which is important, if the load is capacitive.

With a capacitive load, then the LM317 will go into current limiting when the power is first applied, as the load will try to draw a large current. When it's current limiting, the output voltage will be very low, whilst the input is still 100V, causing the maximum rating of 37V between the input and output to be exceeded. This can cause it to be destroyed.

[Zada Moore]

For achieving a fixed low current output of 90V, LM317 can be employed by carefully selecting external resistor values to set the voltage and using a larger output load resistor to limit the current, while ensuring adequate heat dissipation measures are in place.

[marce002]

Ok thanks Zero999 , is clear now...i did not use the diode once i read

"If there are no capacitors on the output, D1 can be omitted,"

since i did not use any cap because there is not in the circuit, i did not use a diode, now i will.

thanks again. no more questions! thks

[Zero999]

Ok thanks Zero999 , is clear now...i did not use the diode once i read

"If there are no capacitors on the output, D1 can be omitted,"

since i did not use any cap because there is not in the circuit, i did not use a diode, no i will.

thanks again. no more questions! thks

Just because you didn't connect a capacitor to the output, it doesn't mean there isn't a capacitor. You didn't say what the load is. Many loads are capactive.

[marce002]

The load is not capacitive it is just a kind of signal to controller (ebike), to fix 90v in the red thin "power switch bat sense line" to tell de controller that the battery remains at 90 vol when 100V are present, and not lower power to motor. if the signal is less than 90V then no problem (means I need to recharge)

My issue here is what i have said before, the output is almost identical to the input less 2 vol , for eg if input is 100v, the output is 98,3V (instead close to 90v),  it is like it follows the input -2v , not fixed at 90v at all.

so i thought was related to the lm317 or missing zener, now i do not know what can be wrong, sorry to bother.

[ejeffrey]

The problem is that your resistor values are off.  The voltage divider ratio is 81:1.  With the 1.25 reference voltage that sets the regulator for 101.25 V.  With your input only at 100V the regulator can't do that so you get Vin-2V.

You need to hit a voltage divider of 72:1. 

[marce002]

hks ejeffrey  will try with 130ohm to achieve your suggested ratio and report back, thks

[Zero999]

The load is not capacitive it is just a kind of signal to controller (ebike), to fix 90v in the red thin "power switch bat sense line" to tell de controller that the battery remains at 90 vol when 100V are present, and not lower power to motor. if the signal is less than 90V then no problem (means I need to recharge)

What do you think is inside it?

There are bound to be some decoupling capacitors.

My issue here is what i have said before, the output is almost identical to the input less 2 vol , for eg if input is 100v, the output is 98,3V (instead close to 90v),  it is like it follows the input -2v , not fixed at 90v at all.

so i thought was related to the lm317 or missing zener, now i do not know what can be wrong, sorry to bother.

Test it at a lower voltage. Connect it to a low voltage PSU, say 5V to 24V. Set it to 1.25V out by connecting the output to the adjust pin, via your 130R resistor and connect the adjust pin to 0V. The output should read close to 1.25V, give or take 50mV.

That will confirm the LM317 isn't fried.

The others who've commented about the output voltage being set to high are also right.

R1 is four 470 resistors in parallel

470/4 = 117.5 Ohms

R2 is two 4k7 resistors in series
4700 + 4700 = 9400 Ohms

V_OUT = (1+R2/R1)*1.25 = (1+9400/117.5)*1.25 = (1+80)*1.25 = 81*1.25 = 101.25V

Actually, it'll be a bit higher, since I didn't take into account I_ADG which is 50µA though R2. V = IR = =0.00005*9400 = 0.47V, so 101.25 + 0.47 = 101.72V, call it 102V.

To summarise, the resistor values you've chosen will give 102V, so the LM317's voltage will not be regulated, until the input exceeds 105V or so and then it'll be 102V, rather than the desired 90V.

[marce002]

thanks a lot, worked perfect. Thanks zero999 and ejeffrey and all, with a value around 125ohm i have perfect constant 89.9V when desired!!! thanks, and no more post.