I find that Bob Cordell's stability test points (Figure 4.9 of 1st edition / Figure 6.9 of 2nd edition) often give early warning signs. And his "stability probe" (Figure 4.12 / Figure 6.12) has helped me immensely. I find that aberrant behavior can be readily tamed using Bob's Zobel network at point P3. Pretty clearly the correct value of Rzobel in Bob's examples, is about 38 ohms.
I don't have much to add in the compensation/stability discussion (except questions 
), but it might help me to understand at least a little bit, if John could could show a schematic with the capacitor positions simulated and discussed?
If I get the overall conclusion right, fast outputs seem to simulate quite well? Was this a 69 bootstrap cap version, or with ccs? When I tried fast transistors with ccs, there was pretty bad oscillation, but it seemed to be related to the driver transistor and upper part of the circuit (upper output and ccs).
With the bootstrapped 69 version, I have not had any oscillation problems with fast transistors in my experiments. This includes PNP and NPN version on different PCB's, and quite few different transistors.
), but it might help me to understand at least a little bit, if John could could show a schematic with the capacitor positions simulated and discussed?If I get the overall conclusion right, fast outputs seem to simulate quite well? Was this a 69 bootstrap cap version, or with ccs? When I tried fast transistors with ccs, there was pretty bad oscillation, but it seemed to be related to the driver transistor and upper part of the circuit (upper output and ccs).
With the bootstrapped 69 version, I have not had any oscillation problems with fast transistors in my experiments. This includes PNP and NPN version on different PCB's, and quite few different transistors.
Last edited:
Self seems to pour cold water onto anything except the collector base VAS capacitor. His reasoning is detailed and not just because it's easy. He shows a simple shunt capacitor to rail that works but takes most of the current. A Zobel maybe. Someone told me to look at Mr Hoods solution in later amps. I'm not sure I ever found it. The JLH isn't a simple VAS so an alternative solution could be helpful.
If my assumption is right fast transistors with additional cog type capacitors might be ideal.
Let me risk some speculation. The open loop gain of this amplifier is so low as to be almost impossible to make it oscillate. As it can I suspect that the division of current in the VAS might be the cause. If so for slightly different reasons the Cdom of the Self amplifiers might be best. Mitsubishi Teleton GA202 in the early 1970s used one in a similar layout although conventional class AB. 100 pF.
If my assumption is right fast transistors with additional cog type capacitors might be ideal.
Let me risk some speculation. The open loop gain of this amplifier is so low as to be almost impossible to make it oscillate. As it can I suspect that the division of current in the VAS might be the cause. If so for slightly different reasons the Cdom of the Self amplifiers might be best. Mitsubishi Teleton GA202 in the early 1970s used one in a similar layout although conventional class AB. 100 pF.
Regarding the circuit, the one I simulated is the original four transistor JLH 69. It is the most stable of the variants. Every additional transistor adds another potential phase shift that may need to be dealt with.
The capacitor in question is added, when used, in parallel with the 2.7k feedback resistor.
Yes, the results here suggest that all combinations are stable, but the very high bandwidth with fast transistors is of concern as even short leads can have enough inductance to cause problems.
Regarding stability test points, I have always measured (at least in simulations, where it is easy to do) the currents in all stages. That can be quite revealing. Perhaps not as detailed as Bob's recommendations which may help.
Single pole dominant gain control is the most stable situation. In the quest for low distortion it is not the best, because it slows down the high frequency response when it is most needed for a fast correction to crossover distortion. Maybe Class A circuits need not be too concerned.
The main other objection is whether stages ahead of the Miller compensation overload. That depends on whether the input signal is fast (generally the industry says not) enough to overload the input until the delayed feedback catches up. If using Miller compensation I degenerate the input stages and run at high enough current not to overload under any signal conditions (as Ed Cherry pointed out and in Stochino's non-slewing designs for example). That seems to be the best of both worlds.
I'm not sure about your low open loop gain suggestion. It does help though but what is probably more significant is that there are only three stages: the input, the driver and the output. That is easier to stabilise than a four or five stage high power Class AB amplifier, I suggest.
The capacitor in question is added, when used, in parallel with the 2.7k feedback resistor.
Yes, the results here suggest that all combinations are stable, but the very high bandwidth with fast transistors is of concern as even short leads can have enough inductance to cause problems.
Regarding stability test points, I have always measured (at least in simulations, where it is easy to do) the currents in all stages. That can be quite revealing. Perhaps not as detailed as Bob's recommendations which may help.
Single pole dominant gain control is the most stable situation. In the quest for low distortion it is not the best, because it slows down the high frequency response when it is most needed for a fast correction to crossover distortion. Maybe Class A circuits need not be too concerned.
The main other objection is whether stages ahead of the Miller compensation overload. That depends on whether the input signal is fast (generally the industry says not) enough to overload the input until the delayed feedback catches up. If using Miller compensation I degenerate the input stages and run at high enough current not to overload under any signal conditions (as Ed Cherry pointed out and in Stochino's non-slewing designs for example). That seems to be the best of both worlds.
I'm not sure about your low open loop gain suggestion. It does help though but what is probably more significant is that there are only three stages: the input, the driver and the output. That is easier to stabilise than a four or five stage high power Class AB amplifier, I suggest.
I use fast output Sanken LAPT transistors 2SC3284 Y (hFE 180), which has an fT around 70MHz at 2A. I also used 2SC5242 (2SC5200) before which has an fT around 30Mhz.
For Q3, I tested about 30-40 different transistors, mainly with low Cob and high hFE and fT.
It only happened to me twice that the JLH69 oscillated. In both cases the outputs were 2SC3284 and for Q3 2SC1941 (NEC) and 2SC3117 which I cannot guarantee is original. I only had a problem with 2SC5242 with 2SC3117. All the transistors I used in the tests were original, the ones I suspected I didn't even solder. (2SC3117 was an exception)
The oscillations can also be seen on an ordinary voltmeter, the offset voltage at the output then constantly goes crazy in the range of +/- 200-300mV and is impossible to adjust.
I am not an electronics technician so I did not go into further analysis of the reasons for oscillations, my assumption is high fT outputs and low Cob Q3.
I solved the problem by adding just 4.7pF MICA in parallel with 2K7 in the feedback.
The sound is best for me with the combination of 2SC1941 and 2SC3284, though it's no worse than 2SC2682 (not oscillating) instead of 2SC1941. This combination has been in operation for about a year now without any speaker protection.
The amplifier is completely stable and sounds great.
In the feedback I dropped the capacitor and corrected the current through Q4 with CCS with fets, for Q3 I have CCS with DN2540.
I made the PCBs myself and they are smaller than the ones from ebay, the outputs are not on the PCB but connected with a couple of cm of wire, there is no compensation for the speakers, and I use speaker cables of 2x4mm2 (bi wiring) 2.5m in length.
I wish someone would try inserting these modifications or at least the transistors I used to see if they would oscillate, especially on ebay PCBs because it may not be a problem at all in the circuit and compensation but in the fake parts and the PCB.
https://www.semicon.sanken-ele.co.jp/sk_content/2sc3284_ds_en.pdf
For Q3, I tested about 30-40 different transistors, mainly with low Cob and high hFE and fT.
It only happened to me twice that the JLH69 oscillated. In both cases the outputs were 2SC3284 and for Q3 2SC1941 (NEC) and 2SC3117 which I cannot guarantee is original. I only had a problem with 2SC5242 with 2SC3117. All the transistors I used in the tests were original, the ones I suspected I didn't even solder. (2SC3117 was an exception)
The oscillations can also be seen on an ordinary voltmeter, the offset voltage at the output then constantly goes crazy in the range of +/- 200-300mV and is impossible to adjust.
I am not an electronics technician so I did not go into further analysis of the reasons for oscillations, my assumption is high fT outputs and low Cob Q3.
I solved the problem by adding just 4.7pF MICA in parallel with 2K7 in the feedback.
The sound is best for me with the combination of 2SC1941 and 2SC3284, though it's no worse than 2SC2682 (not oscillating) instead of 2SC1941. This combination has been in operation for about a year now without any speaker protection.
The amplifier is completely stable and sounds great.
In the feedback I dropped the capacitor and corrected the current through Q4 with CCS with fets, for Q3 I have CCS with DN2540.
I made the PCBs myself and they are smaller than the ones from ebay, the outputs are not on the PCB but connected with a couple of cm of wire, there is no compensation for the speakers, and I use speaker cables of 2x4mm2 (bi wiring) 2.5m in length.
I wish someone would try inserting these modifications or at least the transistors I used to see if they would oscillate, especially on ebay PCBs because it may not be a problem at all in the circuit and compensation but in the fake parts and the PCB.
https://www.semicon.sanken-ele.co.jp/sk_content/2sc3284_ds_en.pdf
I have just been playing with switchmode power supplies. Everything I have tried has been virtually useless except one thing. Reduced voltage. This is totally logical. There was a big difference using 90 VAC over 115 VAC. From 235 VAC down to 90 VAC 50 Hz gave a 16 dB reduction in 50 Hz component of ripple at half load. One piece of equipement didn't mind and another did. The JLH would mind.
In Europe yellow building site transformers are cheap and of excellent quality. I would imagine eBay for £20 is possible. Here is the best bit. As the JLH is class A it's current draw is reasonably constant. To explore 115 VAC down to 90 VAC might be a old filement lamp ( 100 watt 230 V type ? ). You also get to see to current use as light albeit yellow. As these transformers isolate safety is a little better. Most SMPS do 90 VAC for Japan, it's Japan at full load. For European people a 230 to 115 V transformer is useful. Don't get wraped up in the isolation stories. The big deal here is 3 KVA is a nice amount of iron. My Variac worked just as well. If the lamp modulates a lot I suspect the sound will be boomy. At 80 VAC the SMPS might get very hot.
In Europe yellow building site transformers are cheap and of excellent quality. I would imagine eBay for £20 is possible. Here is the best bit. As the JLH is class A it's current draw is reasonably constant. To explore 115 VAC down to 90 VAC might be a old filement lamp ( 100 watt 230 V type ? ). You also get to see to current use as light albeit yellow. As these transformers isolate safety is a little better. Most SMPS do 90 VAC for Japan, it's Japan at full load. For European people a 230 to 115 V transformer is useful. Don't get wraped up in the isolation stories. The big deal here is 3 KVA is a nice amount of iron. My Variac worked just as well. If the lamp modulates a lot I suspect the sound will be boomy. At 80 VAC the SMPS might get very hot.
Stability Issues
With regard to the popular view that high fT transistors are the best option, these defeat the stability mechanism devised by JLH.
Some of these transistors also have far higher current gains than 2N3055. In a Class A system the output transistor will operate at elevated temperatures which will boost the current gain of the devices as compared with specifications based on room temperature 25 degrees C.
Simulations need to take higher temperatures into account.
It is predictable the end result of these combined factors will be an increase in the gain bandwidth product (The gain where the frequency is 1) of the whole amplifier.
Gain will decline with increasing frequency and phase will start to change a decade in frequency below the - 3dB point where the rate of fall off will be 6 dB/Octave or 20 dB/Decade. On a graph that is a 45 degree maximum decline.
Stability depends on a race between phase reaching -180 degrees and gain reducing to 1 the theoretical minimum of a series feedback amplifier.
Unfortunately with a multi stage circuit phase will always get to -180 degrees before gain reduces to 1 unless the voltage is rolled off at a suitable frequency somehow.
A rule of thumb for stability is to arrange for the compensation to take effect a decade in frequency below the -3dB point if you want a phase margin of 45 degrees.
The closed loop gain of the JLH 1969 is 13 because the gain bandwidth product is modest - about 600 from memory, and more of what bulk is available is needed to reduce distortion.
This is somewhat like an aircraft operating at a ceiling of say 200 metres being able to descend at 45 degrees and land on a fixed length flight path and runway.
If the flight ceiling is increased by 10 times there is a problem due to the fixed rate of descent (angle 45 degrees).
Increasing closed loop gain would lengthen the descent and runway somewhat - no-one appears to have tried increasing the closed loop gain.
It is accepted that the inverting input is valid for nfb signals however these are mixed with whatever the speaker throws back and is at risk of picking up any emi in the environment. The inverting input is a summing point for all signals detected.
Capacitor impedance decreases with increasing frequency so a lead capacitor is a short cut path that entails risks for incipient signals to gain back door entry. It would be better to take the capacitor feed from Q3 collector a practice JLH adopted for his other designs.
Just because oscillation is does not manifest itself in obvious ways does not exclude the possibility of unwanted transient effects affecting the sound. If this is extremely bright this is colouration.
With regard to the popular view that high fT transistors are the best option, these defeat the stability mechanism devised by JLH.
Some of these transistors also have far higher current gains than 2N3055. In a Class A system the output transistor will operate at elevated temperatures which will boost the current gain of the devices as compared with specifications based on room temperature 25 degrees C.
Simulations need to take higher temperatures into account.
It is predictable the end result of these combined factors will be an increase in the gain bandwidth product (The gain where the frequency is 1) of the whole amplifier.
Gain will decline with increasing frequency and phase will start to change a decade in frequency below the - 3dB point where the rate of fall off will be 6 dB/Octave or 20 dB/Decade. On a graph that is a 45 degree maximum decline.
Stability depends on a race between phase reaching -180 degrees and gain reducing to 1 the theoretical minimum of a series feedback amplifier.
Unfortunately with a multi stage circuit phase will always get to -180 degrees before gain reduces to 1 unless the voltage is rolled off at a suitable frequency somehow.
A rule of thumb for stability is to arrange for the compensation to take effect a decade in frequency below the -3dB point if you want a phase margin of 45 degrees.
The closed loop gain of the JLH 1969 is 13 because the gain bandwidth product is modest - about 600 from memory, and more of what bulk is available is needed to reduce distortion.
This is somewhat like an aircraft operating at a ceiling of say 200 metres being able to descend at 45 degrees and land on a fixed length flight path and runway.
If the flight ceiling is increased by 10 times there is a problem due to the fixed rate of descent (angle 45 degrees).
Increasing closed loop gain would lengthen the descent and runway somewhat - no-one appears to have tried increasing the closed loop gain.
It is accepted that the inverting input is valid for nfb signals however these are mixed with whatever the speaker throws back and is at risk of picking up any emi in the environment. The inverting input is a summing point for all signals detected.
Capacitor impedance decreases with increasing frequency so a lead capacitor is a short cut path that entails risks for incipient signals to gain back door entry. It would be better to take the capacitor feed from Q3 collector a practice JLH adopted for his other designs.
Just because oscillation is does not manifest itself in obvious ways does not exclude the possibility of unwanted transient effects affecting the sound. If this is extremely bright this is colouration.
JlH printed letters from a fan of his. The next step up from the original transistor types gave better results. The further step didn't sound as good. JLH was surprised and as far as I know never understood exactly why. It's not important what the exact steps were, just to know the designer was at a loss to know all the scientific reasons why things don't always work.
Alternative Devices
This is an interesting suggestion.
The original JLH1969 used MJ480 transistors having fT of 4MHz and good current gain characteristics. The datasheet for MJL21194 shows fT with current in the operating range of 1-2A as being in the region of 7MHz which is more than the nominal 4MHz.
An attraction is with a matched pair of MJL21193/4 THD at 1kHz is 0.08% and 0.8% if unmatched. The spread of tolerances in current gain with modern devices is better than it was in 1969 and we are using identical devices rather than complements here.
The articles referred to in an earlier post about Rod Elliott's website show complementary devices are approximations not mirror images of one another.
THD tests at 1kHz and 20kHz are based on sine wave testing which is unrepresentative of music which consists of transient asymmetric wave forms. The rise times of these can represent higher frequencies which need to be reproduced without significant error. A 10kHz square wave test will show if there is overshoot or ripple in the result and if there is significant delay in settling time.
One can look at Gain bandwidth product as capital in the bank. Some of this needs to be available for forward gain and some for negative feedback.
The apportionment can be changed to improve stability - this by increasing the closed loop gain - which will reduce the frequency where the gain is reduced to 1 (known as unity).
There are some long running threads (Symasym?) where by design closed loop gains of around 50 have been used for Class AB designs using high fT output transistors.
I am running the JLH 1996 which I built in that year using epitaxial base 2N3055's. The closed loop gain is 13 which for my system makes a line stage mandatory.
In the subsequent design to the JLH1969 the JLH1970 15-20W Class AB design covered in the articles on Rod Elliott's website, the closed loop gain is 19 =(2.2k/120)+1.
If stability can be achieved by manipulating the closed loop gain it may be possible to avoid adding a lead capacitor in the feedback loop.
I will leave the experimenting for others to consider. I am spending most of my time listening to music and leaving well enough alone.
This is an interesting suggestion.
The original JLH1969 used MJ480 transistors having fT of 4MHz and good current gain characteristics. The datasheet for MJL21194 shows fT with current in the operating range of 1-2A as being in the region of 7MHz which is more than the nominal 4MHz.
An attraction is with a matched pair of MJL21193/4 THD at 1kHz is 0.08% and 0.8% if unmatched. The spread of tolerances in current gain with modern devices is better than it was in 1969 and we are using identical devices rather than complements here.
The articles referred to in an earlier post about Rod Elliott's website show complementary devices are approximations not mirror images of one another.
THD tests at 1kHz and 20kHz are based on sine wave testing which is unrepresentative of music which consists of transient asymmetric wave forms. The rise times of these can represent higher frequencies which need to be reproduced without significant error. A 10kHz square wave test will show if there is overshoot or ripple in the result and if there is significant delay in settling time.
One can look at Gain bandwidth product as capital in the bank. Some of this needs to be available for forward gain and some for negative feedback.
The apportionment can be changed to improve stability - this by increasing the closed loop gain - which will reduce the frequency where the gain is reduced to 1 (known as unity).
There are some long running threads (Symasym?) where by design closed loop gains of around 50 have been used for Class AB designs using high fT output transistors.
I am running the JLH 1996 which I built in that year using epitaxial base 2N3055's. The closed loop gain is 13 which for my system makes a line stage mandatory.
In the subsequent design to the JLH1969 the JLH1970 15-20W Class AB design covered in the articles on Rod Elliott's website, the closed loop gain is 19 =(2.2k/120)+1.
If stability can be achieved by manipulating the closed loop gain it may be possible to avoid adding a lead capacitor in the feedback loop.
I will leave the experimenting for others to consider. I am spending most of my time listening to music and leaving well enough alone.
That absolutely right. There are other reasons to select the gain. I noticed on this thread an almost passive acceptance that low distortion is desirable. Not if it causes instability. A variation of the Hitachi MOSFET design was unity gain stable ! I built an active filter inside it's feedback loop.This was to drive a 16 inch bass on a four foot baffle. I got 30 Hz flat and some output lower. It's the only amplifier I know of that allows this.
Just been doing a switch mode test. Reducing input voltage to 90 VAC helps as said before. Using ground to 0V is a disaster. We are talking a 40 dB change of fed through ripple best to worse( 15 dB better using 90 VAC ). I wasn't able to get a better result using all conventional filter ideas. I think I tried all common types.
Doublless someone out there knows differently. I would say for a Newby this would be a problem. Personally I would recomend a regulated power supply using LD1085 as heat output can be better than other common types and they seem easy to use. If using switchmode I suspect it needs to float with absolutly no mains earth. I used 100nF 0V to ground with equally bad results. If a laptop PSU behaves differently I couldn't say. I doubt it. I didn't ground the positive as a test. I feel even if it worked I would have doubts.
I didn't leave it there. Grounding 0V or in this case 9V positive does give a nice reduction as long as it's the right place ( at the test oscillator a disaster one metre away). Star ground made very real. However 90 VAC is best and is made worse by any ground. The good news is no oscilloscope required as hum ripple will be heard when wrong. I am using 500 mA. I could have pretended I didn't get the bad result. No point as it can and will happen. It doesn't happen to me often. It did this time. Even with a total ground plane I had it happen. This required a T shaped input copper to the plane to solve it. It was at - 130 dB below 12 V. I was getting -98 dB which was -125 dB ( - 100 dBV ) with the T at 1.5 amps. How did I do that ? I took the failed PCB and made the T out of thick wire. It was an easy gamble to do a new PCB as to be honest the bad one was already good. The sales guy from a company told me they get -140 dB using friends in Formula One's test gear. I doubt they do.
People except Naim Audio see the power supply as something that needs to be the right voltage and right current. Naim amplifiers sounded very different as they grew in size. It was mostly power supply. Julian Vereker told me the cheapest amplifier had the most failed prototypes as the power supply had to be cheaper. He said an amplifier is a signal modulated power supply. The Naim transformers were mildly wrong by choice. This is the surge current on switch on was very high. This would be to others a defect. With a JLH it wouldn't help as that is a class AB or D trick.
Doublless someone out there knows differently. I would say for a Newby this would be a problem. Personally I would recomend a regulated power supply using LD1085 as heat output can be better than other common types and they seem easy to use. If using switchmode I suspect it needs to float with absolutly no mains earth. I used 100nF 0V to ground with equally bad results. If a laptop PSU behaves differently I couldn't say. I doubt it. I didn't ground the positive as a test. I feel even if it worked I would have doubts.
I didn't leave it there. Grounding 0V or in this case 9V positive does give a nice reduction as long as it's the right place ( at the test oscillator a disaster one metre away). Star ground made very real. However 90 VAC is best and is made worse by any ground. The good news is no oscilloscope required as hum ripple will be heard when wrong. I am using 500 mA. I could have pretended I didn't get the bad result. No point as it can and will happen. It doesn't happen to me often. It did this time. Even with a total ground plane I had it happen. This required a T shaped input copper to the plane to solve it. It was at - 130 dB below 12 V. I was getting -98 dB which was -125 dB ( - 100 dBV ) with the T at 1.5 amps. How did I do that ? I took the failed PCB and made the T out of thick wire. It was an easy gamble to do a new PCB as to be honest the bad one was already good. The sales guy from a company told me they get -140 dB using friends in Formula One's test gear. I doubt they do.
People except Naim Audio see the power supply as something that needs to be the right voltage and right current. Naim amplifiers sounded very different as they grew in size. It was mostly power supply. Julian Vereker told me the cheapest amplifier had the most failed prototypes as the power supply had to be cheaper. He said an amplifier is a signal modulated power supply. The Naim transformers were mildly wrong by choice. This is the surge current on switch on was very high. This would be to others a defect. With a JLH it wouldn't help as that is a class AB or D trick.
It's a warning that it might be more complex using a switching power supply. Seeing you have invested so much time a linear power supply seems a logical step.As said LD1084 is the easy route It's not the lowest on dropout but is very stable. TNT Audio do a very good analysis on the lm317 that covers the LD1084. I am very lucky. I seldom have to buy anything as it exists in my scrap pile. It means I never need spend money to try an idea. Very often a daft idea is the best idea. Because I don't have to buy something I just do it.
If TNT make anyone look at TS431 do read the pdf very carefully. On semiconductors from memory has lowest noise. The right capacitor is usually very small or very big In the middle won't work. A fast home made Darlington to complete. Dropout might be an issue. Very low dropout may cause stability problems.
If TNT make anyone look at TS431 do read the pdf very carefully. On semiconductors from memory has lowest noise. The right capacitor is usually very small or very big In the middle won't work. A fast home made Darlington to complete. Dropout might be an issue. Very low dropout may cause stability problems.
I am satisfied with my regulators for JLH69. So I have about 40W of dissipation on voltage regulation and filtering, this is unfortunately a price to pay for a good result.
I tested the regulators at full power with a 1kHz signal at the input of the JLH69, the positive and negative voltages are completely stable and clear with no trace of 1kHz. The reference is LM329, my favorite.
I also planned to test some 3 pin regulators, but this is no longer a priority.
https://www.diyaudio.com/forums/solid-state/3075-jlh-10-watt-class-amplifier-580.html#post6050539
I tested the regulators at full power with a 1kHz signal at the input of the JLH69, the positive and negative voltages are completely stable and clear with no trace of 1kHz. The reference is LM329, my favorite.
I also planned to test some 3 pin regulators, but this is no longer a priority.
https://www.diyaudio.com/forums/solid-state/3075-jlh-10-watt-class-amplifier-580.html#post6050539
I like the ld1084 as it's so easy to use. It has a high current rating and reasonably good dropout voltage. Devices that have better dropout tend to be more fussy.
Why not do what I do. Buy a transformer slightly higher in voltage than required. Then use a Variac to find turn the ideal low voltage.
Why not do what I do. Buy a transformer slightly higher in voltage than required. Then use a Variac to find turn the ideal low voltage.
Not many audio DIYs will have a variac and probably shouldn't, given the intrinsic risk potential of such auto-transformers. In the link below, a few years ago, another Oz DIY bought one of the cheap and popular types but found that active (live) and neutral wiring were transposed Obviously, if you are playing with transformers and making temporary connections, you would be well inside the danger zone with that blunder.
Some of us have years of industry experience and probably a few "bites" to remind us of the need for safe wiring anywhere a connection to the mains is made -whether it is of the live/neutral, ground referenced or floating/balanced type. 'Good for those with experience but others may still be unaware of the dangers in relying on the manufacturers of the cheap auto-transformer type of variac to keep them relatively safe.
Don't do it unless you have had the variac inspected and tested for safety. That may not be a realistic expectation where you live but you could use a test plug if that, at least, is reliable. Otherwise, carefully inspect the wiring connections.
Here's some more:
Variacs are dangerous ? | All About Circuits
Obviously, if you are playing with transformers and making temporary connections, you would be well inside the danger zone with that blunder.Some of us have years of industry experience and probably a few "bites" to remind us of the need for safe wiring anywhere a connection to the mains is made -whether it is of the live/neutral, ground referenced or floating/balanced type. 'Good for those with experience but others may still be unaware of the dangers in relying on the manufacturers of the cheap auto-transformer type of variac to keep them relatively safe.
Don't do it unless you have had the variac inspected and tested for safety. That may not be a realistic expectation where you live but you could use a test plug if that, at least, is reliable. Otherwise, carefully inspect the wiring connections.
Here's some more:
Variacs are dangerous ? | All About Circuits
Ian. With a bit of thought this whole hobby is a no no. The temptation to do it properly always involves risk. We Brits believe Australia is a land of very dangerous animals, snakes, spiders and acts of nature. We don't say not worth the risk. I bet the risks of electricity are better known than spiders.
Another example. I phoned the largest supplier of liquid nitrogen and asked what safety precautions are required.He asked if a well ventilated space was available. A barn was discussed which was ideal. Next he said some thick gloves and an apron could be provided. He then said that the precautions are the same as boiling water. That really made me think. He went on to say water is slightly more dangerous. We all accept that danger. Nitrogen if it needs saying displaces oxygen. One could burn a candle as the flame would shrink before total danger. There is a tap or fawcett that gives boiling water. I really hate that Idea.
We really should only recommend switch mode power supplies if danger is our prime concern. If anyone wants linear power supply advice I will try to remember warnings. I this country old ladies do fit mains plugs as a matter of pride. That's no less dangerous. I was able to know that from a job I did.
I bought my girlfriend a chainsaw for Christmas. It was her request. I really didn't like that Idea
We really should only recommend switch mode power supplies if danger is our prime concern. If anyone wants linear power supply advice I will try to remember warnings. I this country old ladies do fit mains plugs as a matter of pride. That's no less dangerous. I was able to know that from a job I did.
I bought my girlfriend a chainsaw for Christmas. It was her request. I really didn't like that Idea