Before I got too far down the rabbit hole... sanity check time :-) tl;dr: I want to run servos off nonisolated 160 vdc background: I'm updating this: http://www.delorie.com/electronics/bldc/ So the old driver used an external motor power supply (I had a 80VDC linear but it's big and used hard to get parts) and a separate external 15V driver supply (closed box switcher). The servos I've got are designed to run off 160 VDC and the power chip can certainly handle it, which lends itself to non-isolated house voltage (with isolation on the other side of things, of course). I assume this is the target application for these servos. In the USA we have 240v center tapped service. So if I use a simple two diode rectifier, with neutral as 0v, I should be getting 160V relative to "everything else" as neutral is electrically connected to ground (eventually) L1 -->|--*-----*-------- L2 -->|--' | --- 160 VDC --- | N ------------*-------- Since the 15v supply is isolated, it should work just fine with this, but it occurs to me... I don't need isolation here either. However, finding info on an easy 160v to 15v buck regulator design is difficult, and often involves custom wound transformers anyway. the servos have a protective ground on the UVW cable; I assume that goes to earth ground and not neutral. What, if anything, goes between ground and neutral here? I'm thinking EMI. Debugging live is a no-go, but I've got a 160v isolated DC power supply on order for such. It's variable, and enough for debugging, but not enough for full power live usage. Since I need multiple driver boards, I figured I'd make one "power unit" for the 160VDC that had the diodes, capacitors, EMI, breakers, etc. and bring DC to the other boards as if it were a vanilla DC power supply. With the 80v power supply I could use regenerative braking, but that has drawbacks (a loose connection during hard stopping fried one of my boards). I suppose I'll need overvoltage protection and a way to "dump" excess, or math how much capacitance I'll need - then math inrush protection. so... any of the above may be based on invalid assumptions. I may be missing an obvious better solution. There's a non-zero chance of high voltage sparks and magic smoke events. It may be a stupid idea in general, and I should just get the giant switching power supply. I should buy someone else's solution instead because reasons. I could also use pointers to reference designs and known-goods if anyone has done something similar. Discuss :-) Thanks! DJ
160 vdc power rabbit hole...
Started by ●January 31, 2022
Reply by ●January 31, 20222022-01-31
On a sunny day (Sun, 30 Jan 2022 23:38:27 -0500) it happened DJ Delorie <dj@delorie.com> wrote in <xnbkzs34fw.fsf@delorie.com>:> >Before I got too far down the rabbit hole... sanity check time :-) > >tl;dr: I want to run servos off nonisolated 160 vdc > >background: I'm updating this: http://www.delorie.com/electronics/bldc/ > >So the old driver used an external motor power supply (I had a 80VDC >linear but it's big and used hard to get parts) and a separate external >15V driver supply (closed box switcher). The servos I've got are >designed to run off 160 VDC and the power chip can certainly handle it, >which lends itself to non-isolated house voltage (with isolation on the >other side of things, of course). I assume this is the target >application for these servos. > >In the USA we have 240v center tapped service. > >So if I use a simple two diode rectifier, with neutral as 0v, I should >be getting 160V relative to "everything else" as neutral is electrically >connected to ground (eventually) > >L1 -->|--*-----*-------- >L2 -->|--' | > --- 160 VDC > --- > | >N ------------*-------- > > >Since the 15v supply is isolated, it should work just fine with this, >but it occurs to me... I don't need isolation here either. However, >finding info on an easy 160v to 15v buck regulator design is difficult, >and often involves custom wound transformers anyway.Meanwell switchmode power supplies with 15 V DC output at 2.4 A will work with AC or DC input and are dirt cheap https://www.reichelt.nl/nl/nl/schakelvoeding-36w-2-4a-15v-snt-rs-35-15-p137092.html about 14 Euro, is about 16 US dollars I think. ?
Reply by ●January 31, 20222022-01-31
On Monday, 31 January 2022 at 07:19:52 UTC, Jan Panteltje wrote:> Meanwell switchmode power supplies with 15 V DC output at 2.4 A will work with AC or DC input and are dirt cheap > https://www.reichelt.nl/nl/nl/schakelvoeding-36w-2-4a-15v-snt-rs-35-15-p137092.html > about 14 Euro, is about 16 US dollars I think.I have used the 12V version of that supply. They are nicely made and work fine. Meanwell have a USA and a UK website. The data sheets for most of their power supplies are reasonably good and it is often possible to find the technical reports used to support their regulatory submissions, but you may need to search different country websites to find them. John
Reply by ●January 31, 20222022-01-31
Jan Panteltje <pNaonStpealmtje@yahoo.com> writes:> Meanwell switchmode power supplies with 15 V DC output at 2.4 A will > work with AC or DC input and are dirt cheapYup, and I've got one. I just thought a few dollars worth of on-board parts would save space and wiring. In the end, a separate "logic" power supply probably makes sense anyway. We'll see :-)
Reply by ●January 31, 20222022-01-31
On Mon, 31 Jan 2022 07:18:23 GMT, Jan Panteltje <pNaonStpealmtje@yahoo.com> wrote:>On a sunny day (Sun, 30 Jan 2022 23:38:27 -0500) it happened DJ Delorie ><dj@delorie.com> wrote in <xnbkzs34fw.fsf@delorie.com>: > >> >>Before I got too far down the rabbit hole... sanity check time :-) >> >>tl;dr: I want to run servos off nonisolated 160 vdc >> >>background: I'm updating this: http://www.delorie.com/electronics/bldc/ >> >>So the old driver used an external motor power supply (I had a 80VDC >>linear but it's big and used hard to get parts) and a separate external >>15V driver supply (closed box switcher). The servos I've got are >>designed to run off 160 VDC and the power chip can certainly handle it, >>which lends itself to non-isolated house voltage (with isolation on the >>other side of things, of course). I assume this is the target >>application for these servos. >> >>In the USA we have 240v center tapped service. >> >>So if I use a simple two diode rectifier, with neutral as 0v, I should >>be getting 160V relative to "everything else" as neutral is electrically >>connected to ground (eventually) >> >>L1 -->|--*-----*-------- >>L2 -->|--' | >> --- 160 VDC >> --- >> | >>N ------------*-------- >> >> >>Since the 15v supply is isolated, it should work just fine with this, >>but it occurs to me... I don't need isolation here either. However, >>finding info on an easy 160v to 15v buck regulator design is difficult, >>and often involves custom wound transformers anyway. > >Meanwell switchmode power supplies with 15 V DC output at 2.4 A will work with AC or DC input and are dirt cheap > https://www.reichelt.nl/nl/nl/schakelvoeding-36w-2-4a-15v-snt-rs-35-15-p137092.html >about 14 Euro, is about 16 US dollars I think. >?People make tiny pcb mount things that are AC-line to DC supplies, lower power. Like this: https://www.mouser.com/ProductDetail/CUI-Inc/PBO-3C-5?qs=eP2BKZSCXI51S6OuaLJmDQ%3D%3D https://www.mouser.com/ProductDetail/CUI-Inc/PSK-5D-5?qs=DRkmTr78QASBDZkT%252BYmSQQ%3D%3D -- I yam what I yam - Popeye
Reply by ●February 1, 20222022-02-01
DJ Delorie <dj@delorie.com> wrote:> > Before I got too far down the rabbit hole... sanity check time :-) > > tl;dr: I want to run servos off nonisolated 160 vdc > > background: I'm updating this: http://www.delorie.com/electronics/bldc/ > > So the old driver used an external motor power supply (I had a 80VDC > linear but it's big and used hard to get parts) and a separate external > 15V driver supply (closed box switcher). The servos I've got are > designed to run off 160 VDC and the power chip can certainly handle it, > which lends itself to non-isolated house voltage (with isolation on the > other side of things, of course). I assume this is the target > application for these servos. > > In the USA we have 240v center tapped service. > > So if I use a simple two diode rectifier, with neutral as 0v, I should > be getting 160V relative to "everything else" as neutral is electrically > connected to ground (eventually) > > L1 -->|--*-----*-------- > L2 -->|--' | > --- 160 VDC > --- > | > N ------------*--------It's minor, but why bother with providing 240 to the machine in the first place? The losses from a full bridge rectfier will by minimal, and a half rectifier across 240 doesn't even balance the load, so you still need the neutral. There's just no benefit to this design. The fireworks factor goes up with 240, and while low rating double pole breakers exist, they are usually costly and hard to get. Isolation won't kill you and smaller to even couple kVA machine tool transformers can be used for isolation, and don't cost too much, even off ebay. They usually come in 120/240 and sometmimes 480V primaries, all selectable and 120/240 out. This will "soften" the possible fault currents vs. hanging off the power in your shop.
Reply by ●February 1, 20222022-02-01
Cydrome Leader <presence@MUNGEpanix.com> writes:> It's minor, but why bother with providing 240 to the machine in the first place?The spindle needs 240 anyway.> The losses from a full bridge rectfier will by minimal, and a half rectifier > across 240 doesn't even balance the load, so you still need the > neutral.Right, that's the benefit - you still use the neutral, and it's usable as the zero volt point. That means the circuit's "ground" won't be 120v away from earth ground. And it's not quite a half rectifier; it's two half rectifiers on opposite polarities. The result across the caps is the same as a 120v full rectifier.> There's just no benefit to this design. The fireworks factor goes up > with 240,It shouldn't - 240 center tapped means no single wire is more than 120v away from earth ground, It's no more dangerous than a 120v circuit in that respect. Note that in the USA the center tap is grounded, so we have two 120v live conductors of opposite polarity, or 240v across them.> Isolation won't kill you and smaller to even couple kVA machine tool > transformers can be used for isolation, and don't cost too much, even > off ebay. They usually come in 120/240 and sometmimes 480V primaries, > all selectable and 120/240 out. This will "soften" the possible fault > currents vs. hanging off the power in your shop.Yeah, that becomes "power supply" which is what I did in the past, with bigger and bigger transformers... Math-wise, the machine needs 4 servos at a nominal 2A, 8A peak, so 120v * 8 * 4 = 1 KVA minimum, 4 KVA peak? (and yes, there should be a betting pool on how long before the machine destroys itself ;)
Reply by ●February 1, 20222022-02-01
DJ Delorie <dj@delorie.com> wrote:> Cydrome Leader <presence@MUNGEpanix.com> writes: >> It's minor, but why bother with providing 240 to the machine in the first place? > > The spindle needs 240 anyway.fair enough.>> The losses from a full bridge rectfier will by minimal, and a half rectifier >> across 240 doesn't even balance the load, so you still need the >> neutral. > > Right, that's the benefit - you still use the neutral, and it's usable > as the zero volt point. That means the circuit's "ground" won't be 120v > away from earth ground. > > And it's not quite a half rectifier; it's two half rectifiers on > opposite polarities. The result across the caps is the same as a 120v full > rectifier. > >> There's just no benefit to this design. The fireworks factor goes up >> with 240, > > It shouldn't - 240 center tapped means no single wire is more than 120v > away from earth ground, It's no more dangerous than a 120v circuit in > that respect.You still have 240 across the two rectifiers, and the higher potential fault currents. I work with machines and the destruction, even with fuses on 208/240 circuit boards far exceeds anything you'll see with 120. It really is an entirely different game.> Note that in the USA the center tap is grounded, so we have two 120v > live conductors of opposite polarity, or 240v across them. > >> Isolation won't kill you and smaller to even couple kVA machine tool >> transformers can be used for isolation, and don't cost too much, even >> off ebay. They usually come in 120/240 and sometmimes 480V primaries, >> all selectable and 120/240 out. This will "soften" the possible fault >> currents vs. hanging off the power in your shop. > > Yeah, that becomes "power supply" which is what I did in the past, with > bigger and bigger transformers... > > Math-wise, the machine needs 4 servos at a nominal 2A, 8A peak, so 120v > * 8 * 4 = 1 KVA minimum, 4 KVA peak?The rectification will decrease the power factor, so figure higher than that. Sustained full stall current on all axis sounds semi-unlikely unless you have some sort of incredible crash or like switching all directions at once under full load. Use lots of fuses.> (and yes, there should be a betting pool on how long before the machine > destroys itself ;)I hope it's not the wooden one in the photos. It's nicely constructed.
Reply by ●February 1, 20222022-02-01
Cydrome Leader <presence@MUNGEpanix.com> writes:> You still have 240 across the two rectifiers, and the higher potential > fault currents. I work with machines and the destruction, even with fuses > on 208/240 circuit boards far exceeds anything you'll see with 120. It > really is an entirely different game.Good to know. Maybe I'll design the driver boards "as if" they were live (isolated I/O) but use an isolated supply anyway, until I need more power.> The rectification will decrease the power factor, so figure higher than > that. Sustained full stall current on all axis sounds semi-unlikely unless > you have some sort of incredible crash or like switching all directions at > once under full load. Use lots of fuses.The drivers can limit the current in a stall via software, but yeah. Peak power is if all three axes accelerate at the same time. Worst case is a full deceleration of all three axis at once, because of the regenerative braking. My previous driver board didn't have a TVS on the driver board itself, and a loose power connection during hard braking resulted in a spike way outside what the board could handle. I suspect the new software will need to watch for overvoltage, and either fault or at least stop using regen. I might use one of the old boards to *just* switch in braking resistors. And by "resistors" I mean "light bulbs" because who doesn't want a light show? ;-)>> (and yes, there should be a betting pool on how long before the machine >> destroys itself ;) > > I hope it's not the wooden one in the photos. It's nicely constructed.That's the old one, and yes, these servos were on it. It *did* self-destruct once. Drove the spindle right through the side upright. It's feeling much better now :-) The nice thing about DIY is that it's also repair-it-yourself. Or at worst, you build another one ;-)
Reply by ●February 2, 20222022-02-02
DJ Delorie <dj@delorie.com> wrote:> Cydrome Leader <presence@MUNGEpanix.com> writes: >> You still have 240 across the two rectifiers, and the higher potential >> fault currents. I work with machines and the destruction, even with fuses >> on 208/240 circuit boards far exceeds anything you'll see with 120. It >> really is an entirely different game. > > Good to know. Maybe I'll design the driver boards "as if" they were > live (isolated I/O) but use an isolated supply anyway, until I need more > power. > >> The rectification will decrease the power factor, so figure higher than >> that. Sustained full stall current on all axis sounds semi-unlikely unless >> you have some sort of incredible crash or like switching all directions at >> once under full load. Use lots of fuses. > > The drivers can limit the current in a stall via software, but yeah. > Peak power is if all three axes accelerate at the same time. > > Worst case is a full deceleration of all three axis at once, because of > the regenerative braking. My previous driver board didn't have a TVS on > the driver board itself, and a loose power connection during hard > braking resulted in a spike way outside what the board could handle. > > I suspect the new software will need to watch for overvoltage, and > either fault or at least stop using regen. I might use one of the old > boards to *just* switch in braking resistors. And by "resistors" I mean > "light bulbs" because who doesn't want a light show? ;-) > >>> (and yes, there should be a betting pool on how long before the machine >>> destroys itself ;) >> >> I hope it's not the wooden one in the photos. It's nicely constructed. > > That's the old one, and yes, these servos were on it. It *did* > self-destruct once. Drove the spindle right through the side upright. > It's feeling much better now :-) > > The nice thing about DIY is that it's also repair-it-yourself. Or at > worst, you build another one ;-)Chain driven machinery can do some scary stuff when timing is off or something does slip. The forces can be pretty wild. I like how belts will at least slip or break before thing get too nasty.
