hi Guys
i need to power an MCU on 60v supply
unfortunately the HV version on lm317 ot 2596 is not available
a friend suggested to connect 2 lm2596 on series for 60v input & use one output
But it didnt work
as a voltage devider to read the 60v dc on analog pin i have used 100k as R1 & 9,09k as R2
using voltage devider calculator & it works But do you have any ideas if i need to use another values?
Linear regulators can be cascaded under circumstances, in switching regulators the switch is stressed with the full input voltage and must be designed for it. There are various HV switching regulators on the market, but probably not within your reach. Check the selection guides of major manufacturers and see what's available for you.
As you consider a linear regulator (LM317), I presume current consumption of your circuit is relative low. A simple solution would be series zener diode to drop part of the input voltage. Other solution could be a buck converter with discrete switching transistor or a flyback converter.
As for 60 V measurement, you should observe the maximal source impedance specification for ADC input. It's no absolute limit but related to possible accuracy degardation.
Linear regulators can be cascaded under circumstances, in switching regulators the switch is stressed with the full input voltage and must be designed for it. There are various HV switching regulators on the market, but probably not within your reach. Check the selection guides of major manufacturers and see what's available for you.
As you consider a linear regulator (LM317), I presume current consumption of your circuit is relative low. A simple solution would be series zener diode to drop part of the input voltage. Other solution could be a buck converter with discrete switching transistor or a flyback converter.
As for 60 V measurement, you should observe the maximal source impedance specification for ADC input. It's no absolute limit but related to possible accuracy degardation.
There are a few options for regulating a 60V supply. One option is to use a high voltage buck converter, such as the TI LM5160. This will allow you to adjust the output voltage to whatever you need it to be. You can also use a high voltage linear regulator, such as the Murata LDO-HV. This will provide a constant output voltage, but it will not be adjustable. You can also use a combination of high voltage MOSFETs and a low voltage linear regulator, such as the LM317, to step down the voltage to the desired level.
Important information is missing: What load current should it be designed for?
For low currents / low power, one can add a series zener to reduce input voltage. Benefit is, that the voltage is much more stable (vs load current) than with a resistive voltage divider.
So my first idea: Zener, capacitor, step down, capacitor
Important information is missing: What load current should it be designed for?
For low currents / low power, one can add a series zener to reduce input voltage. Benefit is, that the voltage is much more stable (vs load current) than with a resistive voltage divider.
So my first idea: Zener, capacitor, step down, capacitor
hi Klaus
the load is an ATMEGA328p MCU - 16x2lcd - ds1307 RTC moc3021 triac driver & voltage divider for measuring input voltage
but this easy solution will not effect the voltage i need to measure ?
unfortunately i've tried to reduce the input voltage with some 1N4007 diodes on series
as i know this diode drops 0.7 volt but this didnt work
the other issue is the voltage devider for measuring on analog pin
using voltage devider calculator ive used 30k for R1 & 5555 ohm for R2 to measuring 32 volts
then i needed to measure 60v so i re calculated it using the same R1 value 30K & the R2 value is 2727.273 ohms
the first calculated values is working for 12-32 reading but the second values are good for voltage more than 45 but cant measure low voltages accurately
so i tried to increase the R1 value to 100k & calculate the R2 value
But i dont see accuracy for measuring various range of volt
what do you suggest for reading 0-60v ?
im using atmega328 & this is the code
Code:
float in_voltage = 0.0;
float R1 = 30000.0;
float R2 = 5555.556;
float ref_voltage = 5.0;
///////////////////////////////////////// Battery voltage routine
ADCVal = analogRead(batMonPin); // read the voltage on the divider on pin A4
pinVoltage = (ADCVal * ref_voltage) / 1024.0;
if ((timerCurrentSeconds) != tm.Second) {//1 second delay uisng ds1307
batteryVoltageV = pinVoltage / (R2 / (R1 + R2));
}
///////////////////////////////////////// Battery voltage routine END
There are a few options for regulating a 60V supply. One option is to use a high voltage buck converter, such as the TI LM5160. This will allow you to adjust the output voltage to whatever you need it to be. You can also use a high voltage linear regulator, such as the Murata LDO-HV. This will provide a constant output voltage, but it will not be adjustable. You can also use a combination of high voltage MOSFETs and a low voltage linear regulator, such as the LM317, to step down the voltage to the desired level.
unfortunately the HV series of all converters is not available here in the shops
(
You can also use a combination of high voltage MOSFETs and a low voltage linear regulator, such as the LM317, to step down the voltage to the desired level.
unfortunately this solution is more expensive than my device itself
i need to do this with common parts
& the new version of regulators like the HV is not available in the shops here
Linear regulators can be cascaded under circumstances, in switching regulators the switch is stressed with the full input voltage and must be designed for it. There are various HV switching regulators on the market, but probably not within your reach. Check the selection guides of major manufacturers and see what's available for you.
As you consider a linear regulator (LM317), I presume current consumption of your circuit is relative low. A simple solution would be series zener diode to drop part of the input voltage. Other solution could be a buck converter with discrete switching transistor or a flyback converter.
As for 60 V measurement, you should observe the maximal source impedance specification for ADC input. It's no absolute limit but related to possible accuracy degardation.
hi FvM
unfortunately as a friend suggested i've used 2 lm2569 buck converters in series & set the output of one of it to 5v & used it to power the board But everything got exploded so i lost my board & the 2 lm2569 mudules
i dont know why & sure i will not try this again for any reason
the easy solution as Klaus suggested is the solution i will try to test
do you suggest to use the zener diode before the lm317 regulator Or use the zener to regulate 5v from the 60?
thanks
this is the only way i know to use zener diodes
My handheld calculator does not accept text like "ATMEGA.."
--> if you did not yet, then it's high time for you to get an idea of load current. As number.
Now you mix two problems, and it's not clear for me what is what?
Your first line of post#6 talks about the power supply problem, the next about measurement, then supply, supply measurement...and so on
I just did talk about the power supply problem...maybe I already was confused by the mix in post#1
Below is the LTspice sim of a circuit that uses two LM317 to give 5V output from a 60V supply.
The first is set to give 30V out so the voltage is never more than 30V across each LM317 even if the output is shorted.
The LM317's will need to be on heat-sinks, their size depending upon the output load current.
The nominal dissipation will be 30V * load-current for U2 and 25V * load-current for U1.
My handheld calculator does not accept text like "ATMEGA.."
--> if you did not yet, then it's high time for you to get an idea of load current. As number.
Now you mix two problems, and it's not clear for me what is what?
Your first line of post#6 talks about the power supply problem, the next about measurement, then supply, supply measurement...and so on
I just did talk about the power supply problem...maybe I already was confused by the mix in post#1
ok Sir
forget about the power supply problem & lets talk about the measurement
i have this issue because i need to supply power via the same source i need to measure
what do you suggest for measuring 0-60v supply?
actually when i applied external 5v to the board & used the voltage divider i read 59.84 as result
& the second problem is that when regulate the 5v from the same 60v it will decrese this voltage
Below is the LTspice sim of a circuit that uses two LM317 to give 5V output from a 60V supply.
The first is set to give 30V out so the voltage is never more than 30V across each LM317 even if the output is shorted.
The LM317's will need to be on heat-sinks, their size depending upon the output load current.
The nominal dissipation will be 30V * load-current for U2 and 25V * load-current for U1.
What's the input voltage range of interest? 0...60V? ( Is really 0V "of interest", can it never be a tiny bit above 60.00V?)
What is the ADC's input voltage range (decodable)?
What is the ADC's offset error?
What is the ADC's gain error?
What's the ADC's input current / input impedance?
What accuracy and precision do you expect?
What is your sampling rate?
Basically you need a voltage divider to attenuate the input signal.
From your 100k/9.09k it seems you want to step down 60.00V to 5.00V.
This is the right direction .... but for my way of thinking not a good solution, because:
* it limits at (below) 60.00V, so you can never know whether your input is 60.00V, 60.5V, 63.6V.... thus I'd add some overhead
* you think that the ADC decodes up to 5V ... but it decodes always 1 LSB less than VRef. so for a 10 bit ADC it is about 4.995V (max value 1023 of 1024)
* you did not care about ADC offset error and ADC gain error
* you did not care about resistor tolerance
* you did not consider PCB layout problems like GND-bounce: microcontroller_chip_GND vs resistor_divider_GND
* it seems you use VCC as ADC_VRef .... (which I never do for absolute voltage measurements). VCC is a supply voltage, which is not reliable, it is not accurate, nor is it stable. And every error of VRef you will see as 1:1 error in the ADV result. VCC may be +/-10% off, or 5% ... maybe better. It will vary from part to part, drift with time, drift with temperature, drift with load current and it will be noisy.
Thus you need to decide the expected accuracy and precision (my question above)
For example if you want to read 60V with 0.1V (+/-0.05V) accuracy ... then it's a challenge for a 10 bit microcontroller ADC. It is a 60V/0.1V = 600 ratio....
So if your ADC has internal error of 1024LSB/600= 1.67LSB (or +/-0.8 LSB) you will be in trouble
The allowed error is 0.1V/60V = 0.17%.
So all errors in combination need to be smaller than 0.17%, VRef, resistors, ADC...
So it depends on what you need. It may be a simple solution, or it may become a challenge...
In case you are fine that you get a 60V reading with an uncertainty of +/-3V (+/- 5%) then maybe a good Vcc as Vref is suitable.
But in any other case, if you need better accuracy ... you need extra parts, better parts (tolerance) and new calculations.
Again: Thus I asked for your requirements. It's important to give values with units.
If you can't decide values ... then you don't need us .. then just do it in "trial and error" style.
Klaus
--- Updated ---
Power supply:
I don't understand the benefit of using 2 LM317 in series .... vs a voltage drop with a series zener (instead of first LM317 circuit).
Additionally I'd not use a linear supply to step down 60V to 5V with load currents above 10mA or so.
60V x 10mA = 600mW of heat. Continously wasting 11 times the power of the application
Below is the LTspice sim of a circuit that uses two LM317 to give 5V output from a 60V supply.
The first is set to give 30V out so the voltage is never more than 30V across each LM317 even if the output is shorted.
The LM317's will need to be on heat-sinks, their size depending upon the output load current.
The nominal dissipation will be 30V * load-current for U2 and 25V * load-current for U1.
This seems to say input V of first LM317 is limited to 40 V (TI Datasheet) :
Odd its in a table prior to "normal" spec table, and not labeled as worst case. And normal
spec is input to output differential limitation. But one clearly know there has to be a max input
limitation.....
Also sim may have a crappy spice model. eg. sweep input to 100V and see what happens....
hi Klaus
the load is an ATMEGA328p MCU - 16x2lcd - ds1307 RTC moc3021 triac driver & voltage divider for measuring input voltage
but this easy solution will not effect the voltage i need to measure ?
unfortunately i've tried to reduce the input voltage with some 1N4007 diodes on series
as i know this diode drops 0.7 volt but this didnt work
the other issue is the voltage devider for measuring on analog pin
using voltage devider calculator ive used 30k for R1 & 5555 ohm for R2 to measuring 32 volts
then i needed to measure 60v so i re calculated it using the same R1 value 30K & the R2 value is 2727.273 ohms
the first calculated values is working for 12-32 reading but the second values are good for voltage more than 45 but cant measure low voltages accurately
so i tried to increase the R1 value to 100k & calculate the R2 value
But i dont see accuracy for measuring various range of volt
what do you suggest for reading 0-60v ?
im using atmega328 & this is the code
Code:
float in_voltage = 0.0;
float R1 = 30000.0;
float R2 = 5555.556;
float ref_voltage = 5.0;
///////////////////////////////////////// Battery voltage routine
ADCVal = analogRead(batMonPin); // read the voltage on the divider on pin A4
pinVoltage = (ADCVal * ref_voltage) / 1024.0;
if ((timerCurrentSeconds) != tm.Second) {//1 second delay uisng ds1307
batteryVoltageV = pinVoltage / (R2 / (R1 + R2));
}
///////////////////////////////////////// Battery voltage routine END
thanks for help
--- Updated ---
unfortunately the HV series of all converters is not available here in the shops
(
this is what im looking for would you please guide me for this solution
& will this solution effect the input voltage which i need to measure ?
thanks for help
--- Updated ---
unfortunately this solution is more expensive than my device itself
i need to do this with common parts
& the new version of regulators like the HV is not available in the shops here
thanks
--- Updated ---
hi FvM
unfortunately as a friend suggested i've used 2 lm2569 buck converters in series & set the output of one of it to 5v & used it to power the board But everything got exploded so i lost my board & the 2 lm2569 mudules
i dont know why & sure i will not try this again for any reason
the easy solution as Klaus suggested is the solution i will try to test
do you suggest to use the zener diode before the lm317 regulator Or use the zener to regulate 5v from the 60?
* it seems you use VCC as ADC_VRef .... (which I never do for absolute voltage measurements). VCC is a supply voltage, which is not reliable, it is not accurate, nor is it stable. And every error of VRef you will see as 1:1 error in the ADV result. VCC may be +/-10% off, or 5% ... maybe better. It will vary from part to part, drift with time, drift with temperature, drift with load current and it will be noisy.
I have seen a Vref diode used as a shunt regulator for a low power micro Vdd supply to get
an accurate Vref for A/D. Obviously micro cannot have a wide variation in its load current,
but is doable in some designs. One has to pay attention to noise on supply rail as well when
using this approach. Alternatively there are also ways during production test to cal out Vdd
errors and store correction factors in EEPROM :
Lastly there are micros out there with onchip +/- .1% accurate Vrefs now......
Agreed datasheet is incomplete. I took it to be standard test circuit, although
that not called out. And we know that applying 1 KV to the input something not
good is going to happen. This is a question for TI to answer I would think.
I'd start with "how accurate and how clean does this
kooky supply need to be?"
Like maybe a center tapped isolation transformer, a
half wave bridge and some filter caps is good enough?
60V is not a real popular input or output voltage but
you might find adjustable-output telecom-48V DC-DCs
(now 48V is also popular as server intermediate bus)
that could be "bent upward" (or ones which allow you
to use elect-resistor feedback, or kludge in a series R
to the feedback terminal that "makes it so").
Good day. Thank you for contacting TI Customer Support and for choosing our products as part of your solution. This is to acknowledge the support request that you have raised.
The LM317 device is an adjustable three-terminal positive-voltage regulator capable of supplying up to 1.5 A over an output-voltage range of 1.25 V to 37 V. It requires only two external resistors to set the output voltage. The device features a typical line regulation of 0.01% and typical load regulation of 0.1%. It includes current limiting, thermal overload protection, and safe operating area protection. Overload protection remains functional even if the ADJUST terminal is disconnected.
Please see the input voltage range as shown in the table below.
I hope this helps. Please let us know if you need further assistance. Thank you for being a valued customer and we look forward to helping you with your next design.
Kind regards,
Marvin O. Cantos
Texas Instruments Customer Support
Current TI app note (AN-178) is real weak, the old orange
databook (1976) has 3 pages of application figures. However
the only high voltage application circuit uses a vacuum tube
as the voltage dropper, with cathode to VIN, grid to VOUT
and a 40V "zener" for protection (Pg 1-20 of that databook).
I am imagining that this shows how you could use a (beefy
enough) NJFET or depletion mode MOSFET or even a RF
GaN device (those are all depletion mode) if you can find a
high-enough-voltage RF application targeted device.
But there might be a HV regulator from Apex or somebody
like that (their modules are nice, though spendy).
Checked out Pico's line of tiny DC-DCs? They do a lot of
high voltage it appears.
Found a youtube pair of videos where a guy has made
what you describe. Not fan of watching videos myself,
to get to the cheese. But if you freeze frames looks like
a lot of schematic, construction and survivability / safety
is discussed.