Here's an old trick to get a stable reference. Connect a 5.6V Zener in series with a 1A silicon rectifier diode. The result is 6.3V and ends up being a very stable combination. Only an expensive temperature compensated Zener does better. If 6.3V works for you it's a handy thing to know.
True, but you're assuming the heat/temperature is *from* the one, or three, zeners. If the circuit is in a hot / unstable temperature environment, then the variable ambient temperature is what it is.
Also its capacitance decreases. It's widely used for applications where fast dynamics and clippling voltage is required. Adding series diodes decreases the time constant of the circuit.
Reverse biased diodes conduct due to both the Zener effect and the avalanche effect. At low voltage (15 V) the avalanche effect is dominant. But, they have opposite (an unequal) temperature coefficients. At about 6 V (depending on current) the two coefficients exactly compensate each other resulting in zero variation with temperature over a wide temperature range. Data sheets reflect this trend without discussing its causes. Zero x 3 is still zero... We might point out that there are also small resistive surface currents that bypass the junction and creation/recombination currents.
At 00:55 - Putting three devices in series does *not* increase the temperature coefficient by a factor of three - it remains the same - and that is why this arrangement works so well.
555 Zener. Sounds like a great part number for a practical joke. Hook it in series with a transraster, as a buffer for an inverted isolator bank to power your dark emitting diodes.
Could you do a quick video and see if there's anything interesting to see putting both those on the curve tracer? Since a zener has this non-linear thing that happens at the breakdown voltage, I wonder if you can see different behavior as you approach the 15v threshold as the 3 different 5v devices in series each ride over the knee in the curve as compared to a single 15v device?
Low voltage zenners (eg 1V) have a negative temperature coefficient, while high voltage zenners (eg 15V) have a positive temperature coefficient. Arround 5V the temperature coefficient is ZERO. Therefore to minimise variation with temperature you use 5V zenners as your reference, or multiple 5V. See also the wikipedia page on zenner diodes which has a graph of coefficient vs voltage.
At about 5v for a zener diode the temperature coefficient is zero. Above this voltage the coefficient increases positively and bellow about 5v there is a negative temperature coefficient. So choosing three ideally zero temperature coefficient devices in series is better than a 15v zener with a much higher positive temperature coefficient. The temperature coefficient of the three devices will be the same as a single device. Under well matched and ideal case all kept at same temperature. Lower temperature coefficient applied to ( 3 x 5v = 15v combination ).
that's the hard part of design. Temp coefficients for affects in use as temperature changes, changes in values due to ageing is another. Measuring a voltage correctly now is one thing but if you need the same circuit to measure it the same value in 3 years time that's a whole different ball game.
Another advantage of using three 5.1V zeners to make a 15V zener, is that the tolerance of the combination zener will be better than that of a single 15V zener. The combined tolerance of three 5% 5.1V zeners is the square root of [ (Tol1)squared + (Tol2)squared + (Tol3)squared ] or 2.9%.
In theory that would be correct IF the components were randomly sampled from a normal distribution with 5% std deviation - in reality your diodes will most likely be from the same batch and have a common "offset" caused by the current condition/calibration of the manufacturing machine. Also the 5% are usually specified over the complete temperature range and the components are usually with
Interesting experiment. I'd test multiple diodes individually for drift - and then the connections of them too. A fixture made of DIP sockets, some perfboard and ribbon cable could be perfect for that, diodes placed in the socket and all reconfiguration done outside the oven while it is working, allowing changeover while keeping the temperature stable.
@@KeritechElectronics well I mean its just zeners, not exactly a pinnacle of stability or anything, if anything of sorts is needed just use a reference - performance is guaranteed and characterized.
@@VEC7ORlt Reference diodes are all basically constructed that way, especially those used in precision voltage references, where the diodes are selected out of a batch for best performance, then aged while hot, and then selected for zero TC current, and come with a resistor that you use in circuit to drive them with the right current. The reference IC's all are going to drift slightly, so the gold standard is a well aged zener, run with a constant current, at stable temperature in a shielded insulated can, and a buffer that has low noise on the output.
The higher-voltage part is mostly using the avalanche effect, while the low-voltage devices are true zeners. The temperature stability is likely the reason motivating this decision, but another difference is that the avalanche effect has a sharper "knee" so if you throw both on a curve tracer, you'll likely see a much sharper curve for the single-diode arrangement and a much softer one for the three-diode arrangement. In most cases a sharper transition is preferred, but there might be some applications where the more gradual turn-on is desirable.
So interesting - thank You for this idea. But i think it is not only about temperature but also about the dynamic resistance. I would expect a more stable voltage when current changes for a reason or another...
amazing how you can measure temperature using diodes, we made a sensor that had to keep a very stable temperature, this was achieved by also having a diode build in just the right place on the silicon.
On silicon a better sensor comes from making a transistor, as the buried base emitter junction always has lower noise, due to there being no surfaced junction that you can get leakage across the top of the die. Yes most are run with base and collector shorted to make a simple diode, but if you are going to integrate it, a transistor is best and easy, no extra steps in processing involved over what is making the rest. If you are going to integrate the measurement as well, add a little extra die area as well and make a full bandgap reference voltage generator there, as you get as a bonus a stable reference voltage to supply the ADC, and the reference also supplies an amplified die temperature to measure as well, absolutely free. Most DMM chipsets do this, as that temperature reference is used for cold junction compensation for thermocouple use, and the reference is used with the ADC for the voltage range inputs. Use a dual slope converter, needing only 3 external pins for the 2 capacitors, and you can easily get 14 bit accuracy, even on a noisy digital logic chip.
@@SeanBZA the main sensors on the chip were isfet's with current measurements in the pA range. Can't say how the temperature sensing diode was made, to us it was just seen as one on the outside of the sensor. Would have loved to been more involved with the sensor, but the design on silicon was left to a team of semiconductor design experts.
I believe if you were to repeat the test using 3.3 volt zeners in a series chain the temp coefficient would be negative if my cloudy old memory is correct? Again from the muddled up grey matter, I think the zero crossover voltage was around 4.5V? If I recall correctly, forward biased diodes have about -2 mV/deg C or -1.11mV deg F temp coefficient. Love your work and thanks for sharpening up or at least slowing the degeneration of my brain cells. Cheers
I think the surface area would need to be way higher on the 15v zener to compensate for the distributed and discoupled termal dissipation as the 15v zener would create sort of a cloud of heated molecules that would, in ambient pressured, room temperature air move away from the diode but for the little zeners that process would be much quicker...
10mA will lead to an extra 150mW of heat, well within the 250mW heat rating, and after an hour both will be stable in the oven. The temperature inside the oven will not vary, and the die of the 15V zener will be at the same temperature as the dies on the 5V zeners as well, at least within 1C. Those copper leads are suprisingly good at transferring heat away from the die, and diodes are rated for higher temperature than transistors because of this, most are specified up to 150C, while transistors are specified up to 125C die temperature.
@@SeanBZA oh havent thought about heat capacity and conductivity, good point. I mean my theory isnt entirely wrong, its just missing what you said which is the time scale at which it happens. Way quicker than i thought it would. Thanks for taking the time and educating me on this🙏🏼
Hi. Apologizes for asking first, I asked before and lost my post it. what kind of video camera and lens are you using? I bought a amscope mu1003 for my "small components assembly area" that was listed as "high speed" but it isn't high speed enough. As I move the soldering iron or probe into the field of view I am presented with a jerky video. Perhaps the connection between my camera and computer isn't fast enough? Thanks in advance
Really interesting ( never thought of that ! ) ..... FYI , for a handy 8 to 10 volt " Zener " .... you might use the BIG , EHT diode from a microwave oven ... but who knows the temp coefficient ? ( your next task , Heh - Heh ? ) ......... DAVE™🛑
No. On one hand, there is a sweet spot around 5V. On the other hand, where do you find a 1V zener? They start at 2.4V. Occasionally you may find something advertised as a 1V zener, but they are in fact normal diodes with a forward voltage around 1V.
You are using too much current for zener diode. 10.6mA x 15V is about 150mW vs 10.6mA x5V which is about 50mW. At high temperatures without heat sink the devices start to heat up themselves. Three 5V devices have more surface area ( because they are three) to dissipate heat than 15V device. Moreover they have less power to dissipate to start with 50mW vs 150mW. You can try your test using 500uA, that is with 5K ohm series resistor,
The problem there is that the "knee" on low voltage Zeners is poor, so running three 5V Zeners at 500μA will give a high dynamic resistance and hence poor load regulation. For the cost of three Zener diodes, you might as well just use a TL431 and two resistors to get 15V with a typical temperature coefficient of 50ppm/°C and dynamic resistance of 0.2 Ω.
@@plemli Current specified is usually maximum current. It makes no sense to waste 150mW to generate reference voltage. I-V curve of zener diode at beak down is very step. So using less current does not make a significant change to voltage. The resistance of zener diode in break down is almost zero, and your connections and the resistor change contribute to more; though, insignificant change in voltage.
You're exactly wrong here, zeners tend to have very soft knees at low current, 1-10mA is pretty normal current that gets you into the are of low dynamic resistance.
@@aduedc That is wrong on so many different levels. The maximum current is determined from the power dissipation of the diode. I = P/V The datasheet does indeed specify the current required to hit the Zener voltage. Some datasheets will give you a range over which the diode will open. The Zener has a dynamic resistance. This is also on the datasheet. The voltage dropped across a Zener diode is a combination of the Zener voltage AND a product of the dynamic resistance and current. V= Vz + IR Judging from your comment I would not expect you to understand any of this.
Наука
Here's an old trick to get a stable reference. Connect a 5.6V Zener in series with a 1A silicon rectifier diode. The result is 6.3V and ends up being a very stable combination. Only an expensive temperature compensated Zener does better. If 6.3V works for you it's a handy thing to know.
The 3 zener stack also splits the power dissipation between the devices, heating each less.
That was the topic of this video.
True, but you're assuming the heat/temperature is *from* the one, or three, zeners.
If the circuit is in a hot / unstable temperature environment, then the variable ambient temperature is what it is.
Also its capacitance decreases. It's widely used for applications where fast dynamics and clippling voltage is required. Adding series diodes decreases the time constant of the circuit.
Reverse biased diodes conduct due to both the Zener effect and the avalanche effect. At low voltage (15 V) the avalanche effect is dominant. But, they have opposite (an unequal) temperature coefficients. At about 6 V (depending on current) the two coefficients exactly compensate each other resulting in zero variation with temperature over a wide temperature range. Data sheets reflect this trend without discussing its causes. Zero x 3 is still zero... We might point out that there are also small resistive surface currents that bypass the junction and creation/recombination currents.
Thank you. You often provide me with something new to add to the knowledge base. You did it again today.
At 00:55 - Putting three devices in series does *not* increase the temperature coefficient by a factor of three - it remains the same - and that is why this arrangement works so well.
555 Zener.
Sounds like a great part number for a practical joke.
Hook it in series with a transraster, as a buffer for an inverted isolator bank to power your dark emitting diodes.
Could you do a quick video and see if there's anything interesting to see putting both those on the curve tracer? Since a zener has this non-linear thing that happens at the breakdown voltage, I wonder if you can see different behavior as you approach the 15v threshold as the 3 different 5v devices in series each ride over the knee in the curve as compared to a single 15v device?
3 magic words that always make me happy "let's test it".
That I didn't know! Learnt something really useful in this video!
If the tempco is a % or in ppm, it does not need to be 3 times better.
Cool. Really appreciate your videos
And now we want a version with a combination of different zeners. Just like they did in the old days.
Low voltage zenners (eg 1V) have a negative temperature coefficient, while high voltage zenners (eg 15V) have a positive temperature coefficient. Arround 5V the temperature coefficient is ZERO. Therefore to minimise variation with temperature you use 5V zenners as your reference, or multiple 5V. See also the wikipedia page on zenner diodes which has a graph of coefficient vs voltage.
At about 5v for a zener diode the temperature coefficient is zero. Above this voltage the coefficient increases positively and bellow about 5v there is a negative temperature coefficient. So choosing three ideally zero temperature coefficient devices in series is better than a 15v zener with a much higher positive temperature coefficient. The temperature coefficient of the three devices will be the same as a single device. Under well matched and ideal case all kept at same temperature. Lower temperature coefficient applied to ( 3 x 5v = 15v combination ).
I didn't know that HP made "test-ovens" ? LOL!
the improvement is also due to the lower power dissipation in the 5V zener diodes
That looks like a problem that is more properly solved by statistics.
I'd like to see the data on that - test multiple diodes, estimate the distribution.
Interesting, I never considered the temp coefficient But I almost never use Zener diodes. Thanks
that's the hard part of design. Temp coefficients for affects in use as temperature changes, changes in values due to ageing is another. Measuring a voltage correctly now is one thing but if you need the same circuit to measure it the same value in 3 years time that's a whole different ball game.
Nice video. Excellent test. Thx for sharing.
We are expecting more test videos like this , This is great expreiment 😊
Another advantage of using three 5.1V zeners to make a 15V zener, is that the tolerance of the combination zener will be better than that of a single 15V zener. The combined tolerance of three 5% 5.1V zeners is the square root of [ (Tol1)squared + (Tol2)squared + (Tol3)squared ] or 2.9%.
In theory that would be correct IF the components were randomly sampled from a normal distribution with 5% std deviation - in reality your diodes will most likely be from the same batch and have a common "offset" caused by the current condition/calibration of the manufacturing machine. Also the 5% are usually specified over the complete temperature range and the components are usually with
I think 14.3v zener + 14007 diode in series is even better :)
Tank you. Learned something (new to me) today
Interesting experiment. I'd test multiple diodes individually for drift - and then the connections of them too. A fixture made of DIP sockets, some perfboard and ribbon cable could be perfect for that, diodes placed in the socket and all reconfiguration done outside the oven while it is working, allowing changeover while keeping the temperature stable.
With zeners being so meh, why even bother?
@@VEC7ORlt because an observation was made, and a curious mind desires answers :)
@@KeritechElectronics well I mean its just zeners, not exactly a pinnacle of stability or anything, if anything of sorts is needed just use a reference - performance is guaranteed and characterized.
@@VEC7ORlt yes, TL431 or something like this - pretty useful indeed.
@@VEC7ORlt Reference diodes are all basically constructed that way, especially those used in precision voltage references, where the diodes are selected out of a batch for best performance, then aged while hot, and then selected for zero TC current, and come with a resistor that you use in circuit to drive them with the right current. The reference IC's all are going to drift slightly, so the gold standard is a well aged zener, run with a constant current, at stable temperature in a shielded insulated can, and a buffer that has low noise on the output.
The higher-voltage part is mostly using the avalanche effect, while the low-voltage devices are true zeners. The temperature stability is likely the reason motivating this decision, but another difference is that the avalanche effect has a sharper "knee" so if you throw both on a curve tracer, you'll likely see a much sharper curve for the single-diode arrangement and a much softer one for the three-diode arrangement.
In most cases a sharper transition is preferred, but there might be some applications where the more gradual turn-on is desirable.
I am a broken record, WOW, I love the presentations. Thank You, from Bill
I wonder if there is a current that optimizes the drift?
Zero, but that would bring other design challenges. 🙂
So interesting - thank You for this idea. But i think it is not only about temperature but also about the dynamic resistance. I would expect a more stable voltage when current changes for a reason or another...
The result is no surprise. I think it's possible to reach a zero temperature coefficent with e.g. 5-4-6V.
amazing how you can measure temperature using diodes, we made a sensor that had to keep a very stable temperature, this was achieved by also having a diode build in just the right place on the silicon.
On silicon a better sensor comes from making a transistor, as the buried base emitter junction always has lower noise, due to there being no surfaced junction that you can get leakage across the top of the die. Yes most are run with base and collector shorted to make a simple diode, but if you are going to integrate it, a transistor is best and easy, no extra steps in processing involved over what is making the rest. If you are going to integrate the measurement as well, add a little extra die area as well and make a full bandgap reference voltage generator there, as you get as a bonus a stable reference voltage to supply the ADC, and the reference also supplies an amplified die temperature to measure as well, absolutely free. Most DMM chipsets do this, as that temperature reference is used for cold junction compensation for thermocouple use, and the reference is used with the ADC for the voltage range inputs. Use a dual slope converter, needing only 3 external pins for the 2 capacitors, and you can easily get 14 bit accuracy, even on a noisy digital logic chip.
@@SeanBZA the main sensors on the chip were isfet's with current measurements in the pA range. Can't say how the temperature sensing diode was made, to us it was just seen as one on the outside of the sensor. Would have loved to been more involved with the sensor, but the design on silicon was left to a team of semiconductor design experts.
@@TheEmbeddedHobbyist Then it was definitely a diode connected transistor, as that integrates well with FET production, and also buries the junction.
@@SeanBZA thanks, my main concern was the asic that processed the signals.
I believe if you were to repeat the test using 3.3 volt zeners in a series chain the temp coefficient would be negative if my cloudy old memory is correct? Again from the muddled up grey matter, I think the zero crossover voltage was around 4.5V? If I recall correctly, forward biased diodes have about -2 mV/deg C or -1.11mV deg F temp coefficient. Love your work and thanks for sharpening up or at least slowing the degeneration of my brain cells. Cheers
The three zener’s tempcos multiply ? I don‘t think so. When one has 100ppm/°C, the stack will also have 100ppm/°C.
I think the surface area would need to be way higher on the 15v zener to compensate for the distributed and discoupled termal dissipation as the 15v zener would create sort of a cloud of heated molecules that would, in ambient pressured, room temperature air move away from the diode but for the little zeners that process would be much quicker...
10mA will lead to an extra 150mW of heat, well within the 250mW heat rating, and after an hour both will be stable in the oven. The temperature inside the oven will not vary, and the die of the 15V zener will be at the same temperature as the dies on the 5V zeners as well, at least within 1C. Those copper leads are suprisingly good at transferring heat away from the die, and diodes are rated for higher temperature than transistors because of this, most are specified up to 150C, while transistors are specified up to 125C die temperature.
@@SeanBZA oh havent thought about heat capacity and conductivity, good point. I mean my theory isnt entirely wrong, its just missing what you said which is the time scale at which it happens. Way quicker than i thought it would. Thanks for taking the time and educating me on this🙏🏼
Well nowadays zeners are 5.1 or 5.2v iirc
I guess to account for the internal voltage drop
You could run 3 15v in parallel and reduce the heating because you pass lower current per diode
Nice!
Hi. Apologizes for asking first, I asked before and lost my post it. what kind of video camera and lens are you using? I bought a amscope mu1003 for my "small components assembly area" that was listed as "high speed" but it isn't high speed enough. As I move the soldering iron or probe into the field of view I am presented with a jerky video. Perhaps the connection between my camera and computer isn't fast enough? Thanks in advance
Does the oven have an internal fan to equalise air temperature throughout?
Thsnks. What would the result be if you connect some neg and pos tempco zeners in series.
Any better than the 3in series ?
Maybe, but tricky. The temperature coefficient of a 5.1V Zener as very low compared to lower or higher voltage devices.
Really interesting ( never thought of that ! ) ..... FYI , for a handy 8 to 10 volt " Zener " .... you might use the BIG , EHT diode from a microwave oven ... but who knows the temp coefficient ? ( your next task , Heh - Heh ? ) ......... DAVE™🛑
Hang on, Zener, or Zenner?
Would it be better if the circuit was 15 1 volt zeners?
No. On one hand, there is a sweet spot around 5V. On the other hand, where do you find a 1V zener? They start at 2.4V. Occasionally you may find something advertised as a 1V zener, but they are in fact normal diodes with a forward voltage around 1V.
HP Oven? OMG! Horrible Products. 😁
You are using too much current for zener diode. 10.6mA x 15V is about 150mW vs 10.6mA x5V which is about 50mW.
At high temperatures without heat sink the devices start to heat up themselves. Three 5V devices have more surface area ( because they are three) to dissipate heat than 15V device. Moreover they have less power to dissipate to start with 50mW vs 150mW.
You can try your test using 500uA, that is with 5K ohm series resistor,
The problem there is that the "knee" on low voltage Zeners is poor, so running three 5V Zeners at 500μA will give a high dynamic resistance and hence poor load regulation. For the cost of three Zener diodes, you might as well just use a TL431 and two resistors to get 15V with a typical temperature coefficient of 50ppm/°C and dynamic resistance of 0.2 Ω.
The zener voltage is specified in the datasheet for a specific current, typ. in the order of 10mA, and it may be significantly off at 500uA.
@@plemli Current specified is usually maximum current. It makes no sense to waste 150mW to generate reference voltage. I-V curve of zener diode at beak down is very step. So using less current does not make a significant change to voltage. The resistance of zener diode in break down is almost zero, and your connections and the resistor change contribute to more; though, insignificant change in voltage.
You're exactly wrong here, zeners tend to have very soft knees at low current, 1-10mA is pretty normal current that gets you into the are of low dynamic resistance.
@@aduedc That is wrong on so many different levels.
The maximum current is determined from the power dissipation of the diode. I = P/V
The datasheet does indeed specify the current required to hit the Zener voltage. Some datasheets will give you a range over which the diode will open.
The Zener has a dynamic resistance. This is also on the datasheet. The voltage dropped across a Zener diode is a combination of the Zener voltage AND a product of the dynamic resistance and current. V= Vz + IR
Judging from your comment I would not expect you to understand any of this.