SV03 & SV04 diode replacement

[smurfer77]

[edit: note there is an updated graph in a later post]

I have previously posted my findings of SV03 and SV04 didoe substitutes over in Echowar's thread on STV-3H and STV-4H but it seems to be a bit buried there in the middle of that excellent but long thread and I do see the topic popping up now and again. So I apologize for the repetition but I think this deserves its own thread.

Here is a pic of an SV-03 (green boat package - middle). The little guys on the left I call the spider eggs with white dots [EDIT: these are MV103] - also a 3-junction diode maybe with different power handling, but as you will see below the voltage drop is very similar to the green package. On the right is a pair of the SV04 (in brown boat packaging). These were taken out of my Sansui AU-X11. I had thought from reading around that the voltage drop might be similar to the STV diodes of same number of junctions, but they are a bit different. And, when the diodes are used for dropping voltages in bias circuits etc you really want to get it right!

In the figure below Echowar's data has suffix "EW" in legend. My green boat data is from a single diode from my left AU-X11 board (but looks similar to the others I have on hand (I have two AU-X11s). My Brown boat SV03 data is from the average of 2 diodes, but note that variation of Vf was <5%. White-dot spider-egg data is from the average of 3 eggs and variation was <3%. You will see that the SV03 brown boat (~2.1V @10mA) and the STV-03 (~1.8V @10mA) are quite different devices. Like wise the green boat SV04 (~2.8V @10mA) seems different to the STV-4 (~2.3V @10mA). I tried the 1N4148 at first, but we are talking 0.69V per junction, instead of 0.73V @10mA of the 1N4148.

Here is what I found:

I've tried various diodes that claim to have lower than usual forward voltage in the small signal range but the winner is:
SV03 --> 3x 1N4150
SV04 --> 4x 1N4150.

While the 1N4148 and 1N914B were about 0.735V @10mA, the 1N4150 offered about 0.700 and as you can see in the graph above that adds up over 4 junctions. For the 3x and 4x combinations in the graph I used the diodes from my batch of 10 that had the lowest forward voltage, since they were closer to the value I needed, 0.69V. But the variation in the diodes is fairly small and you should be fine if you just use random diodes (For the ten 1N4150 I tested, the forward voltages at 10mA were: 0.698, 0.699, 0.698, 0.701, 0.700, 0.703, 0.700, 0.702, 0.698, 0.700 (V)). The 1N4150 were Vishay brand. I have also measured the SMD version and this will also work and could be made into a neat little package.

 

[smurfer77]

PS. The reason I left a "?" beside the "white-dot egg" in the legend is that i don't really know it is an SV03. The package is very different to the regular green boat but I don't have a schematic for the amp the diode came out of so don't really know what it is. But it sure measures very similar to the SV03 and is certainly a 3 junction diode.

[EDIT: as noted in first post now, these are MV103]

 

[SolderIron]

Looks like result of many hours of work. Go beyond this, I wonder how well do the solution track the original over hot and cold? I am thinking of a 3 axis plot: Volt, current, temp.

 

[Ronito6]

I have previously posted my findings of SV03 and SV04 didoe substitutes over in Echowar's thread on STV-3H and STV-4H but it seems to be a bit buried there in the middle of that excellent but long thread and I do see the topic popping up now and again. So I apologize for the repetition but I think this deserves its own thread.

Here is a pic of an SV-03 (green boat package - middle). The little guys on the left I call the spider eggs with white dots - also a 3-junction diode maybe with different power handling, but as you will see below the voltage drop is very similar to the green package. On the right is a pair of the SV04 (in brown boat packaging). These were taken out of my Sansui AU-X11. I had thought from reading around that the voltage drop might be similar to the STV diodes of same number of junctions, but they are a bit different. And, when the diodes are used for dropping voltages in bias circuits etc you really want to get it right!

In the figure below Echowar's data has suffix "EW" in legend. My green boat data is from a single diode from my left AU-X11 board (but looks similar to the others I have on hand (I have two AU-X11s). My Brown boat SV03 data is from the average of 2 diodes, but note that variation of Vf was <5%. White-dot spider-egg data is from the average of 3 eggs and variation was <3%. You will see that the SV03 brown boat (~2.1V @10mA) and the STV-03 (~1.8V @10mA) are quite different devices. Like wise the green boat SV04 (~2.8V @10mA) seems different to the STV-4 (~2.3V @10mA). I tried the 1N4148 at first, but we are talking 0.69V per junction, instead of 0.73V @10mA of the 1N4148.

Here is what I found:

I've tried various diodes that claim to have lower than usual forward voltage in the small signal range but the winner is:
SV03 --> 3x 1N4150
SV04 --> 4x 1N4150.

While the 1N4148 and 1N914B were about 0.735V @10mA, the 1N4150 offered about 0.700 and as you can see in the graph above that adds up over 4 junctions. For the 3x and 4x combinations in the graph I used the diodes from my batch of 10 that had the lowest forward voltage, since they were closer to the value I needed, 0.69V. But the variation in the diodes is fairly small and you should be fine if you just use random diodes (For the ten 1N4150 I tested, the forward voltages at 10mA were: 0.698, 0.699, 0.698, 0.701, 0.700, 0.703, 0.700, 0.702, 0.698, 0.700 (V)). The 1N4150 were Vishay brand. I have also measured the SMD version and this will also work and could be made into a neat little package.

Very good to know.

Info on good contemporary diode substitution for old is always gold!
(Bust'n rhymes again. Can't help it. There is always a song in my heart. Often a dark one...but a song)

Thank you once again! (Belatedly)

 

[mbz]

Fantastic work!

Is it possible to confirm tracking once the diodes warm up, 3 points should be plenty, say 40C,60C,80C, probably rig up some sort of temp controlled bath.
Appreciate that this is a lot of work.

 

[smurfer77]

Fantastic work!

Is it possible to confirm tracking once the diodes warm up, 3 points should be plenty, say 40C,60C,80C, probably rig up some sort of temp controlled bath.
Appreciate that this is a lot of work.

Hi mbz, someone else asked about temperature dependence. I was thinking they should get so hot in those critical voltage reference in bias circuit positions where typically current is really low??? Anyway, I can definitely look into this (when time allows).

Maybe rather than just fixing environment temperature and measuring Vf, I could vary current and measure resultant temperature (and Vf). It may be interesting to see at what currents the temperature or Vf start going wierd. Thoughts?

 

[mbz]

Certainly not under estimating the amount of work required but think confirmation of resonable temp tracking would fill that need in the bias circuit.
As it stands the 1n4150 is an excellent result, much better than 1n4148 etc... My interest is in the bias circuit so tracking Vf versus temp. Many
bias devices operate at a few mA, 1-5? which is will within maximum limits, so don't think varying current to 10 or 20mA is that meaningful.

 

[smurfer77]

Certainly not under estimating the amount of work required but think confirmation of resonable temp tracking would fill that need in the bias circuit.
As it stands the 1n4150 is an excellent result, much better than 1n4148 etc... My interest is in the bias circuit so tracking Vf versus temp. Many
bias devices operate at a few mA, 1-5? which is will within maximum limits, so don't think varying current to 10 or 20mA is that meaningful.

My thoughts too (on current), and that is why I don't think it will heat up much at those currents. Maybe I'm wrong. Anyway, it will be interesting to check the response. Will look at it when I get time.

 

[mbz]

My thoughts too (on current), and that is why I don't think it will heat up much at those currents. Maybe I'm wrong. Anyway, it will be interesting to check the response. Will look at it when I get time.

Maybe I've misunderstood. The bias current does not cause the bias diode to heat up. The bias diode is deliberately heated by placing it on the heat sink
so it's properties (Vf etc...) track and adjust to the output transistors as they heat up, hopefully keeping the bias constant through the temperature range.
The outputs carry the amps! I'm sure you know this, been through a few misunderstandings this week...

 

[Ronito6]

Maybe I've misunderstood. The bias current does not cause the bias diode to heat up. The bias diode is deliberately heated by placing it on the heat sink
so it's properties (Vf etc...) track and adjust to the output transistors as they heat up, hopefully keeping the bias constant through the temperature range.
The outputs carry the amps! I'm sure you know this, been through a few misunderstandings this week...

I'm getting quite the education here.

Thanks

 

[smurfer77]

Maybe I've misunderstood. The bias current does not cause the bias diode to heat up. The bias diode is deliberately heated by placing it on the heat sink
so it's properties (Vf etc...) track and adjust to the output transistors as they heat up, hopefully keeping the bias constant through the temperature range.
The outputs carry the amps! I'm sure you know this, been through a few misunderstandings this week...

Ahhh i am with you. I was thinking in application of just fixed voltage reference but yeah where the diode is use as a bias servo i see why you want that temp range. I am so used to old sansui amps where they use transistors as the bias servo but i have seen it done with the diodes too like in my Sony VFET amps. Ok. Will look at Vf as a function of heat sink temp with the diode thermally mated. (heating up a heatsink will be no problem )

Ronito, if i recall correctly in the AU-X1 /X11 these didoes were just used as fixed voltage references. bias servo there is done by that transistor on the driver board mounted so that it touches the heat sink .

 

[Ronito6]

The little guys on the left I call the spider eggs with white dots - also a 3-junction diode maybe with different power handling, but as you will see below the voltage drop is very similar to the green package

Those "little guys" are MV-103 after all.

Enjoy!

 

[smurfer77]

Anything stopping using Vbe multipliers (rubber diode) to replace the harder to find voltage reference diodes? It seems we can chose pretty much any voltage reference value we want with appropriate part selection.

 

[Ronito6]

Anything stopping using Vbe multipliers (rubber diode) to replace the harder to find voltage reference diodes? It seems we can chose pretty much any voltage reference value we want with appropriate part selection.

Interesante'

 

[mbz]

Anything stopping using Vbe multipliers (rubber diode) to replace the harder to find voltage reference diodes?

I've no knowledge in this area. Will they provide the same thermal tracking, that is, as the outputs heat up their Vbe drops so need to reduce the "push"
from the bias transistor by also droping its Vbe (usually done by heatsink mounting/heating)

 

[smurfer77]

I've no knowledge in this area. Will they provide the same thermal tracking, that is, as the outputs heat up their Vbe drops so need to reduce the "push"
from the bias transistor by also droping its Vbe (usually done by heatsink mounting/heating)

Yes, they will compensate for temperature changes in bias servo when appropriately attached; that is exactly why Vbe multipliers are used in bias servos commonly. The junction of the transistors tracks with temperature, just like the junction of a diode.

The only thing (possibly) 'new' I suggest I using them to replace difficult to obtain multijunction diodes in such bias circuits, rather than stringing together parts that kind of match and harder to thermally attached in awkward stringed-packages. Pretty sure it will work nicely, with the reference voltage being precisely chosen by resistor values in the divider. Will try in my Sony VFET amps I think...

 

[mbz]

I, and many other members just learnt something, thanks.

 

[Felixc]

In my Luxman L100 class AB output stage, the drift of Vf versus temperature is more important than the closeness of the Vf between the different diodes, as there is a potentiometer to in series to adjust the bias circuit voltage at idle. Still this is relative, as the low-value power resistors placed at the output transistor emittors are meant to avoid a thermal runaway. The effect of the thermal-tracking diodes will minimize the bias current of the output stage, to reduce the wasted heat. In my case, since I changed the output transistors for different ones, the thermal junction drift of those is surely not identical to the older ones, so knowing the Tdrift of the diode is not relevant. All that made several variables to control and test at once. Too much for me. Too much for just one unit to repair. I readjusted the DC offset and the bias current with the new transistors to the required specs, and there it went. The temperature at the heat sink was the same on both channels.

 

[smurfer77]

As the unknow spider egg looking diodes have been pointed out by Ronito6 as being MV103 I've updated the chart for clarity.

So the summary is:

SV03, MV103 --> 3x 1N4150
SV04 --> 4x 1N4150.

Cheers

 

[restorer-john]

I've got a bag full of various NOS STVs if you want some more data points? I have an adjustable precision current source too. (mostly boat shaped Sankens, brown and green and some spider eggs, only a few horseshoes)