Battery powered "benchtop" powersupply

[Simenhs]

Hi, I have started work on a battery powered benchtop powersupply. My current prototype is around 100x70x25mm in size and can deliver 12V - 1A.

Github: https://github.com/Blue-Smoke-Electronics/PBP/wiki
youtube:

I was wondering, is this something any of you would be interested in using if i ever manages to get it to a finished product?
Do you have any use cases i should have in mind when designing next revision? Which features is a most have in a portable powersupply?

[macaba]

A battery powered supply is useful for low-level precision analog electronics for isolation from the mains (i.e. eliminating leakage current) if it is linear (not switch-mode) and well regulated.
A PDF schematic would help us help you come up with ideas for improvement.

[Ian.M]

Why not 15V?  That was the 'classic' upper limit for low voltage bench PSUs due to its historic common use for traditional OPAMP circuits.  Also anyone working on projects for automotive or leisure marine use will want to be able to go up to at least 14.7V, to be able to test them over the voltage range they will see in service.

Also, back to OPAMPs: as your Basic User Manual claims:

With this connector you can charge your pbp. Connect with a usb cable to a charger or to your computer. The usb source need to be able to deliver 500mA witch should be almoust every usb powersource. The pbp can be used while chargning, and having it pluged in will increase battery life when the powerdraw it high. Se chargning symbol for more about charging.

What happens if one stacks two PBPs in series and configures them for symmetrical positive and negative rails, while both are connected to a PC or other shared USB power source?  If that's possible, it would be useful to have some sort of data link that permits operation as a master-slave pair, for ease of adjusting voltage and current symmetrically, + provision for each to limit the other's voltage to maintain symmetry when their own output voltage is reduced by current limiting.

[Simenhs]

A battery powered supply is useful for low-level precision analog electronics for isolation from the mains (i.e. eliminating leakage current) if it is linear (not switch-mode) and well regulated.
A PDF schematic would help us help you come up with ideas for improvement.

In my current design i am using a switch mode supply to take the the voltage down to output voltage + 2V and using a linear regulator for the last 2V. I am hoping that the linear regulator take a way the rippel from the switching regulator. Using just a linear regulator real limits the output power of the powersupply do to heat generation (can not have a large heatsing- not portable). Also when charging the batteries from 5V some switching is going on. It might be possible to have a menu option to turn of the switching regulator and limit power.

Generated a Pdf now, it's a bitt messy, will clean it up later. https://github.com/Blue-Smoke-Electronics/PBP/blob/main/pcb/altium/Powersupply/Powersupply.pdf

[Ian.M]

I am hoping that the linear regulator take a way the rippel from the switching regulator.

*MOST* linear regulator ICs have poor to abysmal HF PSRR. 

[Simenhs]

Why not 15V?  That was the 'classic' upper limit for low voltage bench PSUs due to its historic common use for traditional OPAMP circuits.  Also anyone working on projects for automotive or leisure marine use will want to be able to go up to at least 14.7V, to be able to test them over the voltage range they will see in service.

Agree, I will try to aim for at least 15V in my next revision, think i might be able to go all the way up to 20V, i could also try to aim for 24V, that is used a lot in industrial automation.

Also, back to OPAMPs: as your Basic User Manual claims:

With this connector you can charge your pbp. Connect with a usb cable to a charger or to your computer. The usb source need to be able to deliver 500mA witch should be almoust every usb powersource. The pbp can be used while chargning, and having it pluged in will increase battery life when the powerdraw it high. Se chargning symbol for more about charging.

What happens if one stacks two PBPs in series and configures them for symmetrical positive and negative rails, while both are connected to a PC or other shared USB power source?  If that's possible, it would be useful to have some sort of data link that permits operation as a master-slave pair, for ease of adjusting voltage and current symmetrically, + provision for each to limit the other's voltage to maintain symmetry when their own output voltage is reduced by current limiting.

I do think that having making it possible to connect multiple units in series would be nice. The Usb charger is isolated from the output of the supply so it should be possible. Have only built one unit so has not been able to test that jet. I have not thought about making the outputs symmetrical. Any idea about how to implement a datalink like that?

[macaba]

In my current design i am using a switch mode supply to take the the voltage down to output voltage + 2V and using a linear regulator for the last 2V.

That's understandable, I think it achieves your goals and keeps the size small. It wouldn't be suitable for me and that's not a problem - I'm not the target market.

[NiHaoMike]

*MOST* linear regulator ICs have poor to abysmal HF PSRR.

Some newer ones like the LT3045 are specifically made for use with switchers.

[Terry Bites]

Is that made from play doh?

[Simenhs]

Is that made from play doh?

Yes, and no. I am hoping to someday be able to urn my living from making smal electronics projects like this. I have never done a project on this scale before (excluding dayjobs where I have ben parts of bigger teams). I am doing this as much to learn about how to make a producable and documentet product, as to learn how to build a powersupply. I figure the best way to learn is to try it out, and pretend that i someday will be able to have it ready for sale. 

[MegaVolt]

A very interesting project. What is the battery capacity?

Second channel? I understand that you can connect two sources, but all the charm of compactness and portability is lost. Still, two meals are still needed. Plus, if the channels are isolated, this will increase the possibilities. Or +/- 15V or 30V

[Simenhs]

A very interesting project. What is the battery capacity?

In the current version i am using a 4000mAh 3.7V battery -> (14.8Wh). I am thinking about replacing this with 6 380mAh 3.7V batteries (8.4Wh) in series. This will give around 30 min runtime at maximum output. 

Second channel? I understand that you can connect two sources, but all the charm of compactness and portability is lost. Still, two meals are still needed. Plus, if the channels are isolated, this will increase the possibilities. Or +/- 15V or 30V

I do agree that two channels would be nice, but i do think it would mean that the size, cost and complexity of the device would increase. If i can find a way to include a secund channel without to much hassel for everyone that only need one channel i will definitely do that.   

[Simenhs]

*MOST* linear regulator ICs have poor to abysmal HF PSRR. 

I did some measurements to se the effect of having a linear regulator(LT3080) after the switcher(TPS55288). The measurements is at 5V with 400mA current. Blue line is before linear regulator, yellow is after. Looks like it goes from 80mV to 5mV rippel.

[Ian.M]

24 dB PSRR isn't *good*, but its certainly an improvement not to be sneezed at.   However the LT3080 datasheet claims its good for over 30dB PSRR at 200kHz (see fig. 'Ripple Rejection, Single Supply' on page 7), and if you split the 'In' and 'Vcontrol' pins and thoroughly passive filter the supply to the latter to reduce its ripple by at least an order of magnitude, you should be able to get nearly 50dB PSRR (See fig. 'Ripple Rejection, Dual Supply, IN Pin' on same page and note the hump near 200KHz).

[Simenhs]

Thanks, I will try adding a filter to Vcontrol pin later(do only have  SOT-223 pakage with internaly connected Vin and Vcontroll on hand)

[Simenhs]

SEEMS A LAME IDEA TO ME,IF I NEEDED A POrtable psu i would just use 1 or 2 sla's more power than a pathetic 1 amp!!.

Thanks for your feedback  

Sla = sealed lead acid battery, right?
I agree that a simple battery is a better way to get a lots of current if that is all you care about. Some circuits need a specific voltage and/or current to function correctly, and for these circuits a more advanced power supply with voltage control and current limiter is nice to have.

If you really just want loots of portable (size and weight limited) amps, a li-po or li-ion battery would be better than a sla.

[not1xor1]

SEEMS A LAME IDEA TO ME,IF I NEEDED A POrtable psu i would just use 1 or 2 sla's more power than a pathetic 1 amp!!.

SLA? Much heavier, larger and less durable... unless they are deep cycle (even heavier).

[thm_w]

Thanks for your feedback  

Just ignore that troll.

[gnif]

SEEMS A LAME IDEA TO ME,IF I NEEDED A POrtable psu i would just use 1 or 2 sla's more power than a pathetic 1 amp!!.

 

Settle down child.

[Simenhs]

A PDF schematic would help us help you come up with ideas for improvement.

I made a video where I go trough the schematics.  https://youtu.be/M0-jC2JUDwM

[nightfire]

SEEMS A LAME IDEA TO ME,IF I NEEDED A POrtable psu i would just use 1 or 2 sla's more power than a pathetic 1 amp!!.

SLA? Much heavier, larger and less durable... unless they are deep cycle (even heavier).

On the other side- depending on where and when you want to use them, lead-acid batteries are easier to come by than Lithium based ones. Or maybe someone has to do some project with no recharging possibilites at hand, but would be able to haul some truck battery (lead-acid, 24V) with him... Also depending on the environment, lead based system have possibly lesser risk of going "KABUMM" when somethings goes catastrophically wrong, which might also be of design consideration.

I myself could probably get some small used 12V sealed lead batteries for cheap or free, and with 12V and something up to maybe 7 or 9 still usable Ah in them, thats a lot of bang for the buck in a system.

[Doctorandus_P]

I get completely lost when I try to view your 18 page schematic. (PDF version).

It's got all kind of small isolated sections with no title of what those sections do.
There are also a few pages with partial block diagrams, but I can't see an overview of a block diagram of the whole thing, while that should be the first page of the schematic.

I also do not like the use of the LT3080.
I understand it's sort of attractive because of it's ease of use, but it's too expensive for what it does. It is just a voltage follower.
Combine that with that you've put two of them parallel. This makes it even more expensive, and a "less good fit" for your application.
It also looks like you have a 100mOhm current measurement shunt behind the LT3080 and even an extra schottky diode, and both further reduce the usefulness of the LT3080.

I see you take a feedback voltage directly from the output (R24 and R25) but I do not know where that is going to or what the total control loop is.
I do see three RC combinations on the LT3080 sheet which suggest that it's going to react really slowly to any change of the output voltage or the feedback loop.
Is the final control loop done in software? This would make it even slower.
Using a NPN BJT as a voltage follower and the (already available) opamp to close a control loop around it (and the shunt resistor and schottky diode) seems a better solution to me.

Using such a big color LCD for a battery powered power supply also seems a bad fit, because it uses quite a lot of energy.
I also find the wind blowing into the microphone quite distracting, and combined with volume differences in the voice it makes the video hard to watch.

I also like the multi channel idea, but it does not need much extra complexity. Implement the multi channel by just using two or more of the same power supplies, and then add some option to put them either in series or parallel. An extra plus if the "slave" channels no not even have an LCD (which is a relatively expensive part) and all channels can be set by the same user interface. For mechanical integration, the units could be made stackable. This would need the banana connectors to be on the side though.
If you have stackable units, then a simplified slave unit could also be just an extra and exchangeable battery pack.

This project does remind me of the Miniware MDP-P905 and the MDP-XP Modular Power Supplies.

[Simenhs]

I get completely lost when I try to view your 18 page schematic. (PDF version).

It's got all kind of small isolated sections with no title of what those sections do.
There are also a few pages with partial block diagrams, but I can't see an overview of a block diagram of the whole thing, while that should be the first page of the schematic.

I just updated the schematics. It is no longer hierarchical which mean it is down form 18 pages to 3. Hope it is easier to read now. https://github.com/Blue-Smoke-Electronics/PBP/blob/main/pcb/altium/Powersupply/Powersupply.pdf

I also do not like the use of the LT3080.
I understand it's sort of attractive because of it's ease of use, but it's too expensive for what it does. It is just a voltage follower.
Combine that with that you've put two of them parallel. This makes it even more expensive, and a "less good fit" for your application.
It also looks like you have a 100mOhm current measurement shunt behind the LT3080 and even an extra schottky diode, and both further reduce the usefulness of the LT3080.

I see you take a feedback voltage directly from the output (R24 and R25) but I do not know where that is going to or what the total control loop is.
I do see three RC combinations on the LT3080 sheet which suggest that it's going to react really slowly to any change of the output voltage or the feedback loop.
Is the final control loop done in software? This would make it even slower.
Using a NPN BJT as a voltage follower and the (already available) opamp to close a control loop around it (and the shunt resistor and schottky diode) seems a better solution to me.

I agree that it would be nice if I am able to find a less expensive solution. It would be really nice to have the schottky diode and sens resistor inside control loop and not affecting output voltage.

I am now compensating for the voltage drop over diode and resistor inn software, any yes, it is slow.
 
Just to make sure, this is what you suggest? (see image)

 

How should I protect U1 from overvoltage or reverse voltage when the device is off.

Using such a big color LCD for a battery powered power supply also seems a bad fit, because it uses quite a lot of energy.

The battery can keep the display on 15.4 houers(theoretically), it is not deal braking, but is a significant part of the static power draw. A bigger problem with this large display is the cost, it is almost half the bom. I am thinking about replasing it with somthing like this https://www.adafruit.com/product/618 But I will have to test if I can make the user interface look good on a smaller display. I do like having a large nice looking screen.

I also find the wind blowing into the microphone quite distracting, and combined with volume differences in the voice it makes the video hard to watch.

Sorry about that, this was my first video with voice. Hopefully I can do better next time.

Thanks for grate feedback I will definitely try out some of your suggestions

[Doctorandus_P]

The schematic looks much better now.
On itself there is nothing wrong with drawing a hierarchical schematic, but it has to be done in an intelligent way. For me that means a complete block schematic on the first page, and then further details in the blocks themselves.

One thing I do not like is drawing rectangles around all schematic sections.
These rectangles do not add anything to a schematic. A bit of emptiness around schematic sections Makes the separation of the parts just as clear without adding clutter. Another disadvantage of such rectangles is that they often have to be changed during updates and modifications of the schematic, and this is mostly just wasted time.

 Adding captions to schematic sections is a good idea, but I usually make the letters a bit bigger to pull some attention to them, and this also makes them readable when zoomed out.

The next (also small) thing is the use of wire labels to make connections.
It's always a bit of a struggle to find a balance. If you make too many connections with only labels, you have to search a lot around the schematic (your pdf has searchable text, so that helps). Drawing all wires sometimes is more of a clutter, and it prevents you from using the space on the paper effectively. This is especially clear on the uC page, with the encoders and the "output switch" drawn all above each other and attached with long wires.

----------------------------
This new version of the control loop is indeed what I meant. It is very similar to most power supply designs. However, getting to a good and working design from this start can be quite a challenge. You want the power supply to react quickly to changes, and this can lead to instability. There are already thousands of similar schematics on the 'net. Some from service manuals of professional power supplies, others from cheaply built or badly designed power supplies.

I see you used 1k resistors to filter the encoder signals, but 100k for the switches. 100k does seem a bit much (but it probably still works).
Bourns has made an application note for their encoders, I think it uses 10k resistors all around. (the exact value is not very critical, but 10k limits currents to 500uA (@5V).

I see you use an USB-C connector, but have only a 1W isolated converter to charge the battery.
Implementing full USB-C power delivery may be a bit much, but adding a connector for an external power delivery brick should be easy. This would allow both for quicker charging, and for using the power supply without a battery.

[Simenhs]

The schematic looks much better now.
On itself there is nothing wrong with drawing a hierarchical schematic, but it has to be done in an intelligent way. For me that means a complete block schematic on the first page, and then further details in the blocks themselves.

One thing I do not like is drawing rectangles around all schematic sections.
These rectangles do not add anything to a schematic. A bit of emptiness around schematic sections Makes the separation of the parts just as clear without adding clutter. Another disadvantage of such rectangles is that they often have to be changed during updates and modifications of the schematic, and this is mostly just wasted time.

 Adding captions to schematic sections is a good idea, but I usually make the letters a bit bigger to pull some attention to them, and this also makes them readable when zoomed out.

The next (also small) thing is the use of wire labels to make connections.
It's always a bit of a struggle to find a balance. If you make too many connections with only labels, you have to search a lot around the schematic (your pdf has searchable text, so that helps). Drawing all wires sometimes is more of a clutter, and it prevents you from using the space on the paper effectively. This is especially clear on the uC page, with the encoders and the "output switch" drawn all above each other and attached with long wires.

Tried removing the blocks, that made it look better. I also increased text size and changed some wire connections to label connections.

This new version of the control loop is indeed what I meant. It is very similar to most power supply designs. However, getting to a good and working design from this start can be quite a challenge. You want the power supply to react quickly to changes, and this can lead to instability. There are already thousands of similar schematics on the 'net. Some from service manuals of professional power supplies, others from cheaply built or badly designed power supplies.

Can you point me to one of the good professional ones? They sounds like what I want  

I see you used 1k resistors to filter the encoder signals, but 100k for the switches. 100k does seem a bit much (but it probably still works).
Bourns has made an application note for their encoders, I think it uses 10k resistors all around. (the exact value is not very critical, but 10k limits currents to 500uA (@5V).

I tested out with 1k resistors on switches first, but had problems with bounce. Guess I overcompensated when selecting a 100K. 100K works well, but it is possible to press buttons so fast that it do not register, so I will try with lower resistor values.

I see you use an USB-C connector, but have only a 1W isolated converter to charge the battery.
Implementing full USB-C power delivery may be a bit much, but adding a connector for an external power delivery brick should be easy. This would allow both for quicker charging, and for using the power supply without a battery.

Thanks for spotting that, in my head it was a 2W isolated converter. I will have to replace that.

In theory a connector to a external power brick should be easy, but how do I make sure nobody uses the connector to connect 12V to my li-po battery(poof).

Thanks again for good feedback