On 3 Oct 2015 09:38:39 GMT, Jasen Betts <jasen@xnet.co.nz> wrote:>On 2015-10-03, Piotr Wyderski <peter.pan@neverland.mil> wrote: >> Hello, >> >> I know that there are hordes of nice comparators in SOT-23, >> but just out of curiosity: is it possible to build a small >> voltage (relatively to GND) detector using just one transistor >> and no negative power supplies? If the voltage is >= VBE, i.e. >> 0.7V, then the task is a no-brainer, but how about half of that, >> say, 0.3V? > >you'll need 2 transistors, a PNP emitter follower to raise the voltage >enough for an NPN to turn on. (this is how LM319 includes ground in >the input range)That's it. Two transistors and one resistor make a zero-threshold open-drain comparator.
Small voltage detector
Started by ●October 3, 2015
Reply by ●October 3, 20152015-10-03
Reply by ●October 3, 20152015-10-03
legg wrote:> Voltages outputed from the secondary of a power transformer are > typically bipolarAfter the bridge the transformer's voltage is unipolar and somewhat shifted (by a diode drop if it is a normal silicon Graetz circuit, way smaller if synchronous).> low impedance and large, so your description of the > measurement needing to be made doesn't make much sense.It does, the result indicates which branch of the H bridge should be turned on in a synchronous rectifier. I wondered how would a discrete-only analog controller work if the switch enable threshold is required to be low. My first guess was to clone the input differential pair from an op-amp, but it would be too complex. So I decided to ask the analog gurus here and Jasen's solution needs only 2 transistors, which is much simpler than my idea.> Low power, single supply comparators are avilable in 5 or 6pin > transistor-sized packaging that will operate with inputs at 100s of > millivolts either side of ground.Yes, and they are also integrated in the MCUs, i.e. no external elements (other than a voltage divider perhaps) are needed, so the question is purely theoretical. Best regards, Piotr
Reply by ●October 3, 20152015-10-03
John S wrote:>>> Forgot to add an illustration: >>> >>> VCC >>> + >>> .|. >>> | | >>> | | >>> '-' >>> | | >>> ___ | |/ >>> In -|___|- -+------| >>> |> >>> | >>> | >>> === >>> GND >> >> that's the simplest option afaik. You might improve it by adding a >> series diode in the input, then Rin can be lower. You may also need a >> 2nd diode to protect against Veb >5v. >> >> >> NT > > Good observation.Thanks to all who answered my question, I've learnt several new tricks. :-) Best regards, Piotr
Reply by ●October 3, 20152015-10-03
Boris Mohar wrote:> Germanium transistor?;-) The question is really about simple sub-threshold voltage detection techniques and it on purpose ignores the fact that we live in the 21st century, where the problem has already been solved by cheap solid-state integrated circuits in nice packages. Currently I like the PNP follower idea most -- just three parts and pretty universal. Best regards, Piotr
Reply by ●October 3, 20152015-10-03
On Sat, 3 Oct 2015 20:52:12 +0200, Piotr Wyderski <peter.pan@neverland.mil> wrote:>legg wrote: > >> Voltages outputed from the secondary of a power transformer are >> typically bipolar > >After the bridge the transformer's voltage is unipolar >and somewhat shifted (by a diode drop if it is a normal >silicon Graetz circuit, way smaller if synchronous). > >> low impedance and large, so your description of the >> measurement needing to be made doesn't make much sense. > >It does, the result indicates which branch of the H bridge >should be turned on in a synchronous rectifier. I wondered >how would a discrete-only analog controller work if the >switch enable threshold is required to be low. My first guess >was to clone the input differential pair from an op-amp, >but it would be too complex. So I decided to ask the analog >gurus here and Jasen's solution needs only 2 transistors, >which is much simpler than my idea. >http://www.magma.ca/~legg/20_Years/1994/1994h.html
Reply by ●October 3, 20152015-10-03
On Saturday, 3 October 2015 20:16:15 UTC+1, Piotr Wyderski wrote:> The question is really about simple sub-threshold voltage > detection techniques and it on purpose ignores the fact that > we live in the 21st century, where the problem has already > been solved by cheap solid-state integrated circuits in > nice packages. Currently I like the PNP follower idea most > -- just three parts and pretty universal.discrete is cheaper here, albeit lower spec. NT
Reply by ●October 4, 20152015-10-04
On Sat, 03 Oct 2015 09:42:39 -0400, Boris Mohar <borism_void_@sympatico.ca> wrote:>On Sat, 3 Oct 2015 11:22:08 +0200, Piotr Wyderski <peter.pan@neverland.mil> >wrote: > >>Hello, >> >>I know that there are hordes of nice comparators in SOT-23, >>but just out of curiosity: is it possible to build a small >>voltage (relatively to GND) detector using just one transistor >>and no negative power supplies? If the voltage is >= VBE, i.e. >>0.7V, then the task is a no-brainer, but how about half of that, >>say, 0.3V? >> >> Best reagrds, Piotr > > Germanium transistor?Somebody makes a mosfet with a zero volt threshold. That and a drain pullup is the minimal discrete zero-sense comparator.
Reply by ●October 4, 20152015-10-04
One problem: as the voltage goes down, the required gain goes up (or so one would assume). This is especially important in a comparator (ideally infinite GBW). So you inevitably need more transistors, or better (GaN maybe??). Three works pretty well: http://seventransistorlabs.com/Images/Discrete_Tube_Supply.png The pair needs to be heavily unbalanced to actually include zero (note the 2.2k is supplying about 6mA, while the gain transistor needs maybe 0.2mA to turn on), but that isn't usually a problem for this type of application. Same for the poor turn-off (the gain transistor will take relatively long storage time here, but that actually helps in this case). Tim -- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website: http://seventransistorlabs.com "Piotr Wyderski" wrote in message news:muo6o0$mu6$1@node2.news.atman.pl... Hello, I know that there are hordes of nice comparators in SOT-23, but just out of curiosity: is it possible to build a small voltage (relatively to GND) detector using just one transistor and no negative power supplies? If the voltage is >= VBE, i.e. 0.7V, then the task is a no-brainer, but how about half of that, say, 0.3V? Best reagrds, Piotr
Reply by ●October 4, 20152015-10-04
On Sunday, 4 October 2015 06:10:40 UTC+1, Tim Williams wrote:> "Piotr Wyderski" wrote in message > news:muo6o0$mu6$1@node2.news.atman.pl... > > Hello, > > I know that there are hordes of nice comparators in SOT-23, > but just out of curiosity: is it possible to build a small > voltage (relatively to GND) detector using just one transistor > and no negative power supplies? If the voltage is >= VBE, i.e. > 0.7V, then the task is a no-brainer, but how about half of that, > say, 0.3V? > > Best reagrds, Piotr> One problem: as the voltage goes down, the required gain goes up (or so > one would assume).I don't see how that would be true. Only the gain at & near threshold V matters, and the amount of gain required depends entirely on the circuit the threshold detector drives. Delta V matters, not V.> This is especially important in a comparator (ideally > infinite GBW). > > So you inevitably need more transistors, or better (GaN maybe??).No NT
Reply by ●October 4, 20152015-10-04
On Saturday, October 3, 2015 at 3:24:01 PM UTC-4, legg wrote:> On Sat, 3 Oct 2015 20:52:12 +0200, Piotr Wyderski > <peter.pan@neverland.mil> wrote: > > >legg wrote: > > > >> Voltages outputed from the secondary of a power transformer are > >> typically bipolar > > > >After the bridge the transformer's voltage is unipolar > >and somewhat shifted (by a diode drop if it is a normal > >silicon Graetz circuit, way smaller if synchronous). > > > >> low impedance and large, so your description of the > >> measurement needing to be made doesn't make much sense. > > > >It does, the result indicates which branch of the H bridge > >should be turned on in a synchronous rectifier. I wondered > >how would a discrete-only analog controller work if the > >switch enable threshold is required to be low. My first guess > >was to clone the input differential pair from an op-amp, > >but it would be too complex. So I decided to ask the analog > >gurus here and Jasen's solution needs only 2 transistors, > >which is much simpler than my idea. > > > http://www.magma.ca/~legg/20_Years/1994/1994h.htmlLOL- your title had me thinking of some strange diode invented by an individual named Orring....hmmmkay.
