BS-170 Mosfet as diode clipper

[soupbone]

I'm experimenting using a BS-170 mosfet in place of a 4148.I know that you have to clip off one the legs of the mosfet,but I'm not sure which one?

[Perrow]

http://www.google.se/images?q=bs170+pinout&um=1&ie=UTF-8&source=univ&sa=X&ei=PjZ_TY7JFtCUswah2ZDkBg&ved=0CFIQsAQ&biw=1280&bih=963

[soupbone]

http://www.google.se/images?q=bs170+pinout&um=1&ie=UTF-8&source=univ&sa=X&ei=PjZ_TY7JFtCUswah2ZDkBg&ved=0CFIQsAQ&biw=1280&bih=963

Cool!So,it looks like you would cut the source of the mosfet...is that correct?

[Perrow]

I would say cut the gate, as the diode is connected to the source and drain. But test it with your multimeter.

[danielzink]

Don't cut it - tie it together.

I don't remember where these quotes came from - I've had them for a long time - so apologies to the original posters:

Example #1
G+S - D = normal silicon diode
Correct. This is the intrinsic body diode. Notice that in an N channel, the G+S must be positive with respect to the drain to get the diode to conduct. The anode is the source, the cathode is the drain, and the gate actually plays no part other than being tied to the source. It is a silicon diode with a forward voltage determined by the silicon doping of the semiconductor substrate. Typical Vf is 0.5 to 0.7V when the diode is clearly into conduction. All diodes begin conducting tiny currents well before they nominally turn on.

Notice that if you had a P channel MOSFET, the polarities are reversed and the anode of the diode is the drain and the cathode is the G+S.

Example #2
G+D - S = diode conneted mosfet = very soft knee = as used in the Shaka B.
This is the different one. The gate shorted to drain means that the "diode" doesn't start to conduct until the gate/drain is more than the threshold voltage more positive than the source (positive for N-channels, negative for P channels that is). So the forward voltage of these diodes is usually 1.5 to 3V. The knee of conduction is quite soft, and so diodes using this don't go from full off to full on, they gradually turn on. This happens to be quite good for soft distortion. This is the only FET diode connection worthy of any special study or pursuit IMHO. The others are just too similar to ordinary diodes.

It's this connection that needs the series ordinary diode to keep the body diode (connection 1) from conducting when the signal reverses.
Example #3
D+S - G = low leakage, high res diode
This does not exist for MOSFETs. MOSFET gates are insulated by 20 volts thickness of high purity glass. This connection exists for JFETs, where you are using the gate-channel junction as a diode forward biased, instead of reverse biased as it is in normal JFET operation. The junction is typically lightly doped, low leakage, and will show typical silicon diode drops (0.5-0.7V) as above. The only reason to do this or the connection above is to get a slightly different silicon diode junction, hoping that a diode not normally intended for use as a diode is different a little bit.

OK, I must admit to being completely confounded by the different and apparently conflicting ways I see different FET's being used as clippers in various pedal circuit modifications. Hoping somebody here can explain to me what's going on. Here's why I'm confused:

    * The first example of this I came across was Dano's recommendation to use a pair of IRF510 or 520 MOSFET's as clippers in the Rat circuit. His layout shows the transistors' gate and drain legs tied together, and 1N34A germanium diodes in series with the source legs. I used this setup in a couple of modified Screamers I've built, and it gives a nice smooth & bluesy overdrive tone.
    * In the Mouse v.2.0, Keith has used a pair of BS170 MOSFET's as clippers on the overdrive side of the circuit. As with Dano's configuration, the drain and gate are tied together, but no series diodes are employed. Keith mentioned in another thread that he tried them with the BS170's (the spots for them are present on the Mouse v.2.0 PCB, but are then jumpered per the build instructions), but they sounded "horrible".
    * CallMeRog's Large Beaver Tweaking sticky in the LB forum suggests using a pair of J201 JFET's as clippers, but his directions explicitly state that the drain and source should be tied together, with no mention of a series diode at all. I used this configuration in the Triangle v.1 Beaver I recently built, and it works very well, giving the pedal a noticeably smoother, warmer tone in that clipping mode. I used this setup in place of one of the stock distortion modes on a v.2.0 Mouse I'm building for a customer, and it sounds quite good there, too.
    * Then, I turn to the "Simplemods" WIKI at DIYstompboxes, and about 2/3 down the page, it describes making diodes from FET's by connecting the gate and source together/to ground! That's it--I give up!


So what the heck is going on here? Is there a "right" way to connect the legs on these FET's for best clipping results? What is the function of the series diodes in these setups, and why do they sound good in one clipping application and lousy in the next? If I sound confused, it's because I am....

Any FET guru's care to explain this confusing topic.

There is no right answer.  You can connect them how you want most connections will give you some form of clipping.  Whether a particular connection sounds good to you is another matter.  There are misunderstandings about what is going on with MOSFET that's why you get "bad" (or more precisely unintended) behaviour.

The JFET is easy the G to D or S forms a silicon diode.  You can connect to D, S or both D and S.

The MOSFET clipper has two modes of operation:

The first is  the body diode.  The internal structure of a MOSFET ends up with a silicon diode pointing from the source to the drain.  Ideally you should connect the gate to the source so that when that diode is reversed biased the MOSFET doesn't conduct ie. it behaves as a diode.   

The second is the active MOSFET mode connected as a clipper, here the gate is connected to the drain.  The clip direction is when current flow from D to S.  In this mode the MOSFET doesn't follow the diode exponential law and the turn-on voltage is determined by the MOSFETs threshold voltage Vt - the the gate threshold is reach current flows from D to S.  In other words the MOSFET is behaving as a voltage limiter.    Now, here's the problem some people miss: if you reverse the voltage on this connection the body diode which is pointing from S to D starts to conduct.  So there is different limiting in both directions for a single MOSFET.   If you were to connect two MOSFET in parallel with this connection the overall behaviours ends-up being like two body diodes in parallel.  The MOSFET clipping behaviour is lost because MOSFET limiter has a turn-on voltage above the body diode turn-on voltage and so the other MOSFETs body diode conducts before the MOSFET limiting mode can occur.  If you want the MOSFET limiter behaviour you have to connect another diode in series with each MOSFET to stop the body diode of the other MOSFET interferring (cathode to drain).    You can of course connect diodes in series with the the body diode connection of a MOSFET but that is done for a different reason.

I've also used FETs as diodes by using the Gate as the Anode and the Source as the cathode. When connected back-to-back the resulting clipped waveform has the most perfectly rounded corners you ever seen. However, it didn't sound all that remarkable to me in a circuit. I didn't really spend much time with it though.

Dan

[Galego]

Maybe take a look at the Hermida Zen Drive schematic...

[WGTP]

If you really want to know about Mosfet clipping, here it is. 

http://www.diystompboxes.com/smfforum/index.php?topic=38581.0

[soupbone]

Don't cut it - tie it together.

I don't remember where these quotes came from - I've had them for a long time - so apologies to the original posters:

Example #1
G+S - D = normal silicon diode
Correct. This is the intrinsic body diode. Notice that in an N channel, the G+S must be positive with respect to the drain to get the diode to conduct. The anode is the source, the cathode is the drain, and the gate actually plays no part other than being tied to the source. It is a silicon diode with a forward voltage determined by the silicon doping of the semiconductor substrate. Typical Vf is 0.5 to 0.7V when the diode is clearly into conduction. All diodes begin conducting tiny currents well before they nominally turn on.

Notice that if you had a P channel MOSFET, the polarities are reversed and the anode of the diode is the drain and the cathode is the G+S.

Example #2
G+D - S = diode conneted mosfet = very soft knee = as used in the Shaka B.
This is the different one. The gate shorted to drain means that the "diode" doesn't start to conduct until the gate/drain is more than the threshold voltage more positive than the source (positive for N-channels, negative for P channels that is). So the forward voltage of these diodes is usually 1.5 to 3V. The knee of conduction is quite soft, and so diodes using this don't go from full off to full on, they gradually turn on. This happens to be quite good for soft distortion. This is the only FET diode connection worthy of any special study or pursuit IMHO. The others are just too similar to ordinary diodes.

It's this connection that needs the series ordinary diode to keep the body diode (connection 1) from conducting when the signal reverses.
Example #3
D+S - G = low leakage, high res diode
This does not exist for MOSFETs. MOSFET gates are insulated by 20 volts thickness of high purity glass. This connection exists for JFETs, where you are using the gate-channel junction as a diode forward biased, instead of reverse biased as it is in normal JFET operation. The junction is typically lightly doped, low leakage, and will show typical silicon diode drops (0.5-0.7V) as above. The only reason to do this or the connection above is to get a slightly different silicon diode junction, hoping that a diode not normally intended for use as a diode is different a little bit.

OK, I must admit to being completely confounded by the different and apparently conflicting ways I see different FET's being used as clippers in various pedal circuit modifications. Hoping somebody here can explain to me what's going on. Here's why I'm confused:

   * The first example of this I came across was Dano's recommendation to use a pair of IRF510 or 520 MOSFET's as clippers in the Rat circuit. His layout shows the transistors' gate and drain legs tied together, and 1N34A germanium diodes in series with the source legs. I used this setup in a couple of modified Screamers I've built, and it gives a nice smooth & bluesy overdrive tone.
   * In the Mouse v.2.0, Keith has used a pair of BS170 MOSFET's as clippers on the overdrive side of the circuit. As with Dano's configuration, the drain and gate are tied together, but no series diodes are employed. Keith mentioned in another thread that he tried them with the BS170's (the spots for them are present on the Mouse v.2.0 PCB, but are then jumpered per the build instructions), but they sounded "horrible".
   * CallMeRog's Large Beaver Tweaking sticky in the LB forum suggests using a pair of J201 JFET's as clippers, but his directions explicitly state that the drain and source should be tied together, with no mention of a series diode at all. I used this configuration in the Triangle v.1 Beaver I recently built, and it works very well, giving the pedal a noticeably smoother, warmer tone in that clipping mode. I used this setup in place of one of the stock distortion modes on a v.2.0 Mouse I'm building for a customer, and it sounds quite good there, too.
   * Then, I turn to the "Simplemods" WIKI at DIYstompboxes, and about 2/3 down the page, it describes making diodes from FET's by connecting the gate and source together/to ground! That's it--I give up!


So what the heck is going on here? Is there a "right" way to connect the legs on these FET's for best clipping results? What is the function of the series diodes in these setups, and why do they sound good in one clipping application and lousy in the next? If I sound confused, it's because I am....

Any FET guru's care to explain this confusing topic.

There is no right answer.  You can connect them how you want most connections will give you some form of clipping.  Whether a particular connection sounds good to you is another matter.  There are misunderstandings about what is going on with MOSFET that's why you get "bad" (or more precisely unintended) behaviour.

The JFET is easy the G to D or S forms a silicon diode.  You can connect to D, S or both D and S.

The MOSFET clipper has two modes of operation:

The first is  the body diode.  The internal structure of a MOSFET ends up with a silicon diode pointing from the source to the drain.  Ideally you should connect the gate to the source so that when that diode is reversed biased the MOSFET doesn't conduct ie. it behaves as a diode.  

The second is the active MOSFET mode connected as a clipper, here the gate is connected to the drain.  The clip direction is when current flow from D to S.  In this mode the MOSFET doesn't follow the diode exponential law and the turn-on voltage is determined by the MOSFETs threshold voltage Vt - the the gate threshold is reach current flows from D to S.  In other words the MOSFET is behaving as a voltage limiter.    Now, here's the problem some people miss: if you reverse the voltage on this connection the body diode which is pointing from S to D starts to conduct.  So there is different limiting in both directions for a single MOSFET.   If you were to connect two MOSFET in parallel with this connection the overall behaviours ends-up being like two body diodes in parallel.  The MOSFET clipping behaviour is lost because MOSFET limiter has a turn-on voltage above the body diode turn-on voltage and so the other MOSFETs body diode conducts before the MOSFET limiting mode can occur.  If you want the MOSFET limiter behaviour you have to connect another diode in series with each MOSFET to stop the body diode of the other MOSFET interferring (cathode to drain).    You can of course connect diodes in series with the the body diode connection of a MOSFET but that is done for a different reason.

I've also used FETs as diodes by using the Gate as the Anode and the Source as the cathode. When connected back-to-back the resulting clipped waveform has the most perfectly rounded corners you ever seen. However, it didn't sound all that remarkable to me in a circuit. I didn't really spend much time with it though.

Dan

Sorry Daniel for sounding like an idiot,But when you say tie together,are you meaning solder together?Is a BS-170 in parallel with a 1n34a a good combo?I'm looking for something that sounds very "Tube-Like".You mentioned an IRF10.Would that be a better combo?isthat mosfet the same pinout as the BS-170?(sorry for all the questions)

[R.G.]

There are two ways to get "diode connected out of a MOSFET.

One is to use the substrate diode, which always allows current to flow backwards to the way it flows from drain to source. This diode is a normal silicon diode characteristic, consisting as it does of just the substrate isolation tubs.

The second is the one we usually want from MOSFETs. It consists of connecting the gate to the drain, and then using the drain-source as the diode in the normal current flow direction. Connecting the control node to the incoming power node (as in base to collector in bipolars) is the way to get the "diode characteristic" of the control node. It's done in bipolars to get a more-ideal diode characteristic than you get from either of the normal junctions. But this is the connection you want: gate to drain, then drain and source as the diode. Don't clip any leads off the MOSFET.

There is a problem here; the reverse/substrate diode prevents this "diode" from blocking in the reverse direction. In one direction it electronically looks like the amplified-diode MOSFET you want, and in the other it's an ordinary silicon diode. You can use an external diode in series with the drain or source to block the substrate diode from conducting, or if you're using two MOSFETs, you can connect them so the substrate diodes themselves are what allow current flow in the "right" directions each way.

[Brossman]

@ R.G. - So then, what are the advantages of such a setup?  Other than a tonal difference (which is very subjective) why would one use such a setup as opposed to regular Si/Ge clipping diodes?

[R.G.]

The diode connected MOSFET is different in kind as well as size.

A BS170 or 2N7000 will have a diode voltage of maybe 3-4V, which will be very difficult to determine because the onset of clipping is not a well defined level, but rather a sloppy, slushy bending over over a wide range. This is a Good Thing for soft distortion. There's a wide range of distortion where it goes slowly from clean to a little muddy to muddier. It's a Bad Thing for metal distortion because it never gets to a full flat-topped clipping unless you really work at it.

[WGTP]

R.G. and John Greene et. al. graciously covered most of this in the link I posted.   

AMZ also has some info and schematics of the the issues being discussed. 

http://www.muzique.com/lab/zenmos.htm

I put together several distortions to utilize the actual Mosfet or switch.   

http://www.aronnelson.com/gallery/main.php/v/WGTP/Soft_Knee.jpg.html?g2_imageViewsIndex=1

http://www.aronnelson.com/gallery/main.php/v/WGTP/Moster.jpg.html?g2_imageViewsIndex=1

[soupbone]

There are two ways to get "diode connected out of a MOSFET.

One is to use the substrate diode, which always allows current to flow backwards to the way it flows from drain to source. This diode is a normal silicon diode characteristic, consisting as it does of just the substrate isolation tubs.

The second is the one we usually want from MOSFETs. It consists of connecting the gate to the drain, and then using the drain-source as the diode in the normal current flow direction. Connecting the control node to the incoming power node (as in base to collector in bipolars) is the way to get the "diode characteristic" of the control node. It's done in bipolars to get a more-ideal diode characteristic than you get from either of the normal junctions. But this is the connection you want: gate to drain, then drain and source as the diode. Don't clip any leads off the MOSFET.

There is a problem here; the reverse/substrate diode prevents this "diode" from blocking in the reverse direction. In one direction it electronically looks like the amplified-diode MOSFET you want, and in the other it's an ordinary silicon diode. You can use an external diode in series with the drain or source to block the substrate diode from conducting, or if you're using two MOSFETs, you can connect them so the substrate diodes themselves are what allow current flow in the "right" directions each way.

Hey R.G.,so your saying on the second way,is to solder the mosfet to the gate and drain together?and,then solder the other end with the drain-source?let's say you were going to to use a mosfet in a simple pedal like a Boss DS-1,where one of the diode markings,or arrow is facing to the "right".I'm confused on which way it would face?Which side of the mosfet would you use?The rounded side,or the flat side?

[cpm]

i also like the asymetrical clipping of only one Mosfet (diode junction and body together)

[R.G.]

Hey R.G.,so your saying on the second way,is to solder the mosfet to the gate and drain together?and,then solder the other end with the drain-source?let's say you were going to to use a mosfet in a simple pedal like a Boss DS-1,where one of the diode markings,or arrow is facing to the "right".I'm confused on which way it would face?Which side of the mosfet would you use?The rounded side,or the flat side?

Let's first consider just the N-channel MOSFET. This is normally used with the drain more positive than the source, current flowing into the drain and out of the source.

First of all, the drain can be made way more positive than the source if there is no voltage on the gate; this is normal "off" operation, and how a MOSFET blocks DC when it's off. When you raise the gate voltage, at some point it starts making the channel conduct a little, then more and more as it goes higher. Again, normal operation. If you connect the gate to the drain, and then raise the voltage on the drain, the MOSFET again does not conduct at first, until the voltage on the gate (and drain it's soldered to) get to the threshold voltage, at which point the gate voltage starts turning the channel conduction on. The only difference with the gate connected to the drain is that the drain soon conducts so much current that it eats all the current whatever is supplying the voltage to it can produce. The transconductance (i.e. drain current per change of gate voltage) is so high that the voltage across drain and source can't rise much more than Vthreshold. This makes for a "diode" which has a big turn-on voltage of about Vthreshold.

The body diode is reverse biased all this time. Unless the voltage from drain to source gets bigger than the breakdown voltage of the MOSFET and body diode, it never conducts at all when the drain is more positive than the source.

But if you reverse the voltage across the MOSFET, making the source more positive than the drain, this can only go on for about one silicon-diode drop before the substrate/body diode starts to conduct from source to drain pins.

The net of this is that a typical MOSFET like the BS170 and 2N7000 is already a back-to-back clipping diode pair if you just connect the gate to the drain. One way it conducts from source to drain, and looks like an ordinary silicon diode; that's the substrate diode. The other way it conducts from drain to source, and is the "MOSFET diode" caused by the gate making the channel conduct.

If you want an asymmetrical diode clipper, you get that by just connecting gate to drain and using drain and source as your diode terminals.

I'm dancing around saying drain, source and gate instead of pins because it is possible that every MOSFET will have a different pinout. You should always think of the terminal functions, and only at the very last worry about what the package/pin arrangement is, because it may be anything. Look it up when you get ready to put solder on wire.

If you want a symmetrical MOSFET diode clipper, you have to figure some way to keep the reverse substrate diode from being what conducts in the reverse direction. This usually takes the form of another diode to allow current to flow in the drain and out the source, but prevent it from flowing the other way.

[soupbone]

There are two ways to get "diode connected out of a MOSFET.

One is to use the substrate diode, which always allows current to flow backwards to the way it flows from drain to source. This diode is a normal silicon diode characteristic, consisting as it does of just the substrate isolation tubs.

The second is the one we usually want from MOSFETs. It consists of connecting the gate to the drain, and then using the drain-source as the diode in the normal current flow direction. Connecting the control node to the incoming power node (as in base to collector in bipolars) is the way to get the "diode characteristic" of the control node. It's done in bipolars to get a more-ideal diode characteristic than you get from either of the normal junctions. But this is the connection you want: gate to drain, then drain and source as the diode. Don't clip any leads off the MOSFET.

There is a problem here; the reverse/substrate diode prevents this "diode" from blocking in the reverse direction. In one direction it electronically looks like the amplified-diode MOSFET you want, and in the other it's an ordinary silicon diode. You can use an external diode in series with the drain or source to block the substrate diode from conducting, or if you're using two MOSFETs, you can connect them so the substrate diodes themselves are what allow current flow in the "right" directions each way.

Hey R.G.,so your saying on the second way,is to solder the mosfet to the gate and drain together?and,then solder the other end with the drain-source?let's say you were going to to use a mosfet in a simple pedal like a Boss DS-1,where one of the diode markings,or arrow is facing to the "right".I'm confused on which way it would face?Which side of the mosfet would you use?The rounded side,or the flat side?

Awesome!!Thanks R.G. for the ideas!

[Steben]

In the OCD one MOSfet does not have a series germ diode? I don't see the point of the circuit actually... since the body diode of that one conducts earlier than the reverse MOSFET + germ.