I agree. One of my best-serving amplifiers is a 50W stereo using 2N3055/MJ2955 in a single ended old fashioned 2000 uF output capacitor before these were considered small. It runs from a linear regulator with 90V input and a 2N3773, serving both channels. Absolutely solid performance. Can't hear hum at all. Not used so much now other than for testing other devices as my main amp is now a 100W using modern HF linear gain transistors.
Seeing as there is the scope to push PSU a little here goes. The replacement of the 1K for 150R was not a great move. This works against feelings one might have. Notice -103dBV hiss. That's exactly like a LM7812 or 7824. In fact a 7812 needs post regulator filtering to do that.
What some might not see is we must keep the voltage down to keep things cool.
This was mostly to show people what you guess is not really good enough. I notice many at DIY ( not so much here ) are very confident they don't need to measure things. No one knows everything and testing things however unlikely is worth the trouble.
I will try a Darlington next. This is using 9 and 12 VAC transformers and a variac. 227 VAC ( 12 V ) gives me 15 VDC on load. 27 VDC or whatever will be much the same. I have found some 22 000 uF 35 VDC when we get to that.
John, it is with transformer regulated supply. Supply voltage looks okay. I tried to bypass regulators and the noise did not disappear. So it must be somewhere in amplifier circuit, which is classic 1969. I use 2SC5200 matched with Hfe about 170 at 1,5A. First transistors are 2SA970, second 2SC2240, all with Hfe around 400. There was some oscillation, so I had to add 47pF across the feedback resistor and also add some 10pF to first and second transistors.
I think there might be a problem with earth circuit imperfection. Will be trying different options. I experience difficulties finding out the best way to do grounding and avoid ground loop from RCA input connectors. Traditionally I am used to connect all earthes per channel in one point from which cables go to chassis where they connect between channels. If I do like this however, the JLH "complains" when I plug in 2 cables connected to a source device due to ground loop. Any tips how to do earthing in JLH is the best way?
The interesting fact though, that after a few minutes the noise disappears completely, the amp then is dead quiet.
Transistors 2SC2240 I bought not from a proven source and they look a bit suspicious. I found a pair of authentic 2SC2320 from an old amp. The specs look good, but the voltage rate is 50V. Will they work instead of 2240?
I think there might be a problem with earth circuit imperfection. Will be trying different options. I experience difficulties finding out the best way to do grounding and avoid ground loop from RCA input connectors. Traditionally I am used to connect all earthes per channel in one point from which cables go to chassis where they connect between channels. If I do like this however, the JLH "complains" when I plug in 2 cables connected to a source device due to ground loop. Any tips how to do earthing in JLH is the best way?
The interesting fact though, that after a few minutes the noise disappears completely, the amp then is dead quiet.
Transistors 2SC2240 I bought not from a proven source and they look a bit suspicious. I found a pair of authentic 2SC2320 from an old amp. The specs look good, but the voltage rate is 50V. Will they work instead of 2240?
The 2SC2240 is a transistor for low frequency and low noise applications. This device is designed to lower noise figure in the region of low signal source impedance, and to lower the pulse noise. This is recommended for the first stages of Equalizer amplifiers.
Collector power dissipation 300mW
not for VAS or driver
What is the base current of the 2SC5200? For 1.5A it will need at least 10mA, 2SC2240 is too weak for that and maybe it is working at the limit now. That may be the problem. Try with BD139 or better with 2SC2682, 2SC3423, 2SD438
Collector power dissipation 300mW
not for VAS or driver
What is the base current of the 2SC5200? For 1.5A it will need at least 10mA, 2SC2240 is too weak for that and maybe it is working at the limit now. That may be the problem. Try with BD139 or better with 2SC2682, 2SC3423, 2SD438
http://raylectronics.nl/pdfs/2SC2240_TOS.pdf
this is too much for 2sc2240
it is not a problem to measure the current through 2SC2240, measure the current through R1 and divide by two,measure the voltage at C and E and you easily get the power that must meet the table from the PDF
the problem is when the 2SC5200 is cold and 2SC2240 can burn out which can cause more damage
this is too much for 2sc2240
it is not a problem to measure the current through 2SC2240, measure the current through R1 and divide by two,measure the voltage at C and E and you easily get the power that must meet the table from the PDF
the problem is when the 2SC5200 is cold and 2SC2240 can burn out which can cause more damage
Last edited:
I wanted to check someting TNT Audio said without them giving detail of how to do it. That is to use a bypassed zener and LM317. Maybe they have a point. Having moved house I can't find things so forgive it's not JLH voltage. I assume 240R default value.
I was to try a Darlington version of post 6524. It wasn't great.
With a better ground the 50 Hz should vanish even with 22 000 uF. I recently built a PSU for a friend. I just removed all the photographic layer off of a PCB blank. Everything was dead bug. I used a fixed voltage regulator for simplicity. It worked very well. It was wanted in a hurry so I had to do what I knew would work. I have built this project the way I suspect people visiting here will do. Hum could be 20 dB better. 34mV -80dB = -109 dB if ideal.
2 x LM317 at 1.2 A 27 V would suit a JLH very well. The 1000 uF seems to be OK on the zener. 100 uF should be fine. This aids noise and output impedance.
Still why not design uLDO? Less heat, less heatsink so less large.
LM317 also is a dinosaur surpassed by better LDO regulators. I recently helped a guy with JLH fed by 2 x SMPS. As expected no hum at all. My thought was to also try out an ideal rectifier based on LT4320 (had excellent results with that), CLC filter and a 24V capable LDO or a discrete LDO design with MOSFET.
I do use such designs on other low power amplifiers with very good results but those aren't so power hungry as class A designs.
LM317 also is a dinosaur surpassed by better LDO regulators. I recently helped a guy with JLH fed by 2 x SMPS. As expected no hum at all. My thought was to also try out an ideal rectifier based on LT4320 (had excellent results with that), CLC filter and a 24V capable LDO or a discrete LDO design with MOSFET.
I do use such designs on other low power amplifiers with very good results but those aren't so power hungry as class A designs.
Last edited:
If your uLDO regulates positive DC output voltage only, there are simpler ways to achieve the same result. Without needing a voltage doubler and without needing a center tapped transformer.
Depletion mode N-channel power MOSFETs accomplish the same function while operating with Vgs < 0V. No above-the-rail voltage doubler needed. These MOSFETs are available with current ratings far exceeding the LM317.
example
Depletion mode N-channel power MOSFETs accomplish the same function while operating with Vgs < 0V. No above-the-rail voltage doubler needed. These MOSFETs are available with current ratings far exceeding the LM317.
example
I always use LM317 as an indicator of where a better device might be better. I would like to think when the ground better arranged I might use 10 000 uF. This means 1V extra headroom if avoiding ripple intrusion. 20 000 uF 35V is a common type. Using generic LD1084 we might get 3.5 safe amps. That requires nearly 2V headroom. That might mean 30V is possible and certainly 27 VDC. If the voltage could be tweaked it could be set by ear. LD1084 is a slightly lower dropout type.It never seems unstable which is prime. If my 34mV rms ripple snap shot is right a 20 watt halogen lamp draws 1 amp at 11.2V. No wonder LM 317 had enough current. 90 000 uF is unusual. RMS is unhelpful for regulator calculations, peak to peak better. People say LM 317 sounds good. I suspect this may be due to being easy to use.
You might have to allow 5 to 7 watts excess heat per channel using LM 317.
You might have to allow 5 to 7 watts excess heat per channel using LM 317.
I tried a 2N3055 with zener and was very impressed with the hiss, Hum was -60dBV using more uF than this case.
Here a LM317 becomes a zener with it's own supply and RC input filter to the LM317. Just a scrap box special. No negative feedback used on the 2N3055 except it's mode of use. I was very surprised that LM317 is doing a very good job, 3055 base a nice LM317 load. It also shows how a current amplifier works. Raw DC in the collector and clean DC from the emitter. The hiss is way better than book ( -90 dBV ?). Only 10 000uF use on the 2N3055. We can infer -80dBV hum at 3.5 amps which is very good. Add 3VDC to raw supply if so. That should improve to circa -86dBV if using 22 000uF ( Add only 1.5 VDC ). That's a guesssing game but is ball park. I am surprised I got this result as the circuit is complex, or in a way it isn't. LD1084 can equal this.
These are my shorthand notes if you like, not really for showing other people so please forgive. I really object to simple 3 terminal devices being redrawn to suit convention and NE555 timers also. One can use a PNP transistor and 10R resistor. This is a marginally better idea as it is very stable. Remember this is an amplifier colouring the actual one. This is very simple so should have a good sound. Use a fast transistor if you like. No real need as the final 10MF is holding a lot of energy. The filter is 159 Hz meaning it comes into use as Dolby B would at 1 kHz ( Hope I got my maths right ).
Note the volt drop between 3055 base and emitter. This will change with load. All parts as cheap as possible as long as the ripple current OK. Modern cheap parts are very good. 63V capacitors often better than others. Reform them if they have been used at lower voltages and need to re-live at higher voltages.
As I said modern parts are mostly very good. Russia had the same ideas about MIG aircraft. Only a complete fool would doubt them. I have been around Soviet engineers who can make things works with basic materials. In that circuit a 2N3055 is probalbly best of the best. It can stand 5 amps base current so won't mind what the LM317 can do. The Russians make steel that can routinely cope with -40C on railways. That's almost impossible. Engineering isn't about spending money. It's about spending enough money. A second supply for the LM317 at 6VA for example.
I liken some of what I see as bathing in Evian water. I notice often people think a special part will be better without saying why it is special. That's not science. The 2N3055 is very low noise most likely because the base will withstand current ( Rbb-? ). Where I live in Shaftesbury the water is as good as Evian so I do bathe is similar water! Our area supplied water.
To better a switchmode ( SMPS ) is harder than people believe. I think with a little thought 20 dB hiss and 10 dB hum improvement over SMPS is possible without being too complicated. I think the last example very reliable. Spend money on uF not make. Switch modes also have a different quality to the noise, it sounds brighter and OK.
I liken some of what I see as bathing in Evian water. I notice often people think a special part will be better without saying why it is special. That's not science. The 2N3055 is very low noise most likely because the base will withstand current ( Rbb-? ). Where I live in Shaftesbury the water is as good as Evian so I do bathe is similar water! Our area supplied water.
To better a switchmode ( SMPS ) is harder than people believe. I think with a little thought 20 dB hiss and 10 dB hum improvement over SMPS is possible without being too complicated. I think the last example very reliable. Spend money on uF not make. Switch modes also have a different quality to the noise, it sounds brighter and OK.
I received a Private Message asking me to please elaborate upon these remarks, and to please design an example circuit which illustrates them.
Figure 1 below is the "uLDO" schematic linked in post #6513 of this thread.
Figure 2 is an alternative implementation using a depletion mode MOSFET. I've added gate oxide protection diodes D1 and D2, which is conservative, standard practice a/k/a "good hygiene." As long as the regulated output voltage exceeds the transistor's threshold voltage {denoted Vgs_off in the IXYS datasheet} plus another 2.5V for the TL431, the circuit works a treat. So, in this case, as long as we require the voltage regulator to produce an output greater than or equal to +7.0 volts, all is well.
_
Thinking it might be best to involve the entire diyAudio membership, I'm putting the schematic here on the Forum, so that all can participate and learn. Anyone is free to offer a comment or a critique or a suggestion. Or to ask a question.
Figure 1 below is the "uLDO" schematic linked in post #6513 of this thread.
Figure 2 is an alternative implementation using a depletion mode MOSFET. I've added gate oxide protection diodes D1 and D2, which is conservative, standard practice a/k/a "good hygiene." As long as the regulated output voltage exceeds the transistor's threshold voltage {denoted Vgs_off in the IXYS datasheet} plus another 2.5V for the TL431, the circuit works a treat. So, in this case, as long as we require the voltage regulator to produce an output greater than or equal to +7.0 volts, all is well.
_
Attachments
Last edited:
Thanks for elaborating. The short circuit protection of the old circuit could be useful here I guess. Since most transformers have double secondary windings a dual mono PSU can be built by duplicating the circuit.
I will try the suggested circuit with IXTP6N50D2 (6A version of the depletion N channel MOSFET) when I am home again. The older circuit outperforms many 3 pin regulators.
I will try the suggested circuit with IXTP6N50D2 (6A version of the depletion N channel MOSFET) when I am home again. The older circuit outperforms many 3 pin regulators.
I found fuses a good protection device. They are not easy choices. Audiophiles like oversized transformers. Here is a good excuse to use one. They seldom cost a lot more. You can crowbar a lm317 as a way of suggesting all is not well. If you take my 0RI output use it to drive a comparator. LM324 makes a good one. Use the spare op amps to show current. It might sound ironic to say fuses are good. They in some ways do the job in a better way. If you have the crowbar you seldom blow the fuse. I would double the VA of the transformer over what the fuse suggests. We ran a 10 amp kettle at work at 6.3 AT fusing for months. We gave up waiting for it to blow. We had at least 20 staff.