The Mellotron has only one channel, but there is a tape head for each key and there are about 2 1/2 octaves- about 30 tape heads, all in series. So any head has to make voltage across the entire head array. Naturally it is going to be noisy. The tape equalizer uses bipolar transistors- with a fairly low input impedance as these things go (think: late-1960s consumer electronics design). I suspect that while it can be made a little quieter by replacing the ancient transistors in the circuit, that with a little redesign that it might be made a little quieter yet- perhaps with an FET input? Its obvious that the EQ amp does not have a lot of gain- it appears to only have two gain stages. The real issue is I have no idea what they set up for EQ. At this point it seems that I'll just have to take it apart and trace it out. Once I have done so I will post the schematic. That will happen *after* this Thursday- I have a gig with it Thursday night.
I'm assuming that any good NAB equalizer preamp would be good.
As they used the same technology that any tape machine uses.
I read that john lennon would record his own sample tapes on an open reel deck at the studio and install them in his mellotron at home.
And that an adapter was eventualy created with guides for 1/4" tape.
The reason I had asked if you had tried replacing the tapes (before I realized that there was'nt a record function) is that back 1990 3m came out with the 996 tape and was virtualy noiseless.
It was what saved the analog recording industery until the 24 bit format became available.
The formula was sold to BASF and now I don't what brand it is under now.
That was when ampex came out with the 499 formula which is comparable but I prefer the 996 as it was better except for some small clicks inherent in the tape problably due to improper mill grinding.
They could only be barley heard in the blank passages they could not be erased and disappeared with the DBX on.
I no longer need to use the DBX function at all on my MSR-16 except to increase the dynamic range and even that is marginaly useful.
as it is much cleaner sounding without it.
I first started using 996 on my FOSTEX R8 1/4" 8-track and what world of difference it was compared to the ampex 457 and 456 that everybody swore was the best.
I still have a few reels a 226 that they used way back then and for the life of me I still don't know how they got by with it.
Besides my tascams I also have two OTARI's the MTR12 and MTR15.
I will compile the plaback headamp circuits to send them to you and then you could go from there. jer
As they used the same technology that any tape machine uses.
I read that john lennon would record his own sample tapes on an open reel deck at the studio and install them in his mellotron at home.
And that an adapter was eventualy created with guides for 1/4" tape.
The reason I had asked if you had tried replacing the tapes (before I realized that there was'nt a record function) is that back 1990 3m came out with the 996 tape and was virtualy noiseless.
It was what saved the analog recording industery until the 24 bit format became available.
The formula was sold to BASF and now I don't what brand it is under now.
That was when ampex came out with the 499 formula which is comparable but I prefer the 996 as it was better except for some small clicks inherent in the tape problably due to improper mill grinding.
They could only be barley heard in the blank passages they could not be erased and disappeared with the DBX on.
I no longer need to use the DBX function at all on my MSR-16 except to increase the dynamic range and even that is marginaly useful.
as it is much cleaner sounding without it.
I first started using 996 on my FOSTEX R8 1/4" 8-track and what world of difference it was compared to the ampex 457 and 456 that everybody swore was the best.
I still have a few reels a 226 that they used way back then and for the life of me I still don't know how they got by with it.
Besides my tascams I also have two OTARI's the MTR12 and MTR15.
I will compile the plaback headamp circuits to send them to you and then you could go from there. jer
The tape formulation in the Mellotron is not much of an issue. The noise that concerns me has nothing to do with the noise of the tape- it is everything about the noise made when the machine is just sitting. Its before the volume control, but I don't have to play a single note and the noise if the EQ preamp is quite evident.
Quite often the input impedance of a preamp and the source impedance can have an interaction that results in the unit making more noise than it has to. In this case, I suspect that part of the problem is that the input impedance is rather low while the combined series impedance of the heads is rather high. That is why I have been thinking of an FET input- if I could create a high input impedance it might be that a lot of the preamp noise will vanish.
I highly doubt that the transistors have become noisy. I suspect the hiss is by design, so to speak. The tape heads output a fairly low signal and I think they're fairly high impedance as well. That's about the worst case for noise. Combine that with the fact that you have a lot of channels running in parallel being summed at the end would make matters worse.
I would vote for reducing the noise bandwidth if you can. But that would probably require a design change or tweak.
~Tom
I would suggest the opposite, the heads are likely to be fairly low impedance, and the preamp designed accordingly - but in any case, increasing the input impedance of the preamp it more likely to increase noise than reduce it.
You say the noise is there all the time, have you tried shorting all the heads out (same as shorting the only head in a normal tape player), what does that do to the noise?.
Hmm. The heads are typical of 1/2 track mono heads designed for 1/4 inch tape. They are on a special track that can be shifted so you can pick up the 3 different tracks on the 3/8" wide tape. Since there are about 30 heads in series, even if they had a fairly low impedance (for a late 60s design), it seems to me the resulting impedance will still be quite high.
I've not tried shorting the input yet. That seems like it would tell us a lot about what is going on.
So there is only one preamp and the tape head combines the signal from the differently pitched tapes? There might be an easy way to mute the entire circuit till a key is hit. As far as sound quality, if that tape is more than 10 to 15 years old and the thing was used (this tape gets used a lot more than normal tape machine usage (how many times has the middle C been used, 1000's ? )) there may not be any "quality" left on that tape. Also, when was the last time tape head was aligned and the repro electronics calibrated? (although Im not sure how you would do this without an alignment tape or I guess pieces of tape?)
I rebuilt a good portion of the instrument when I obtained it about 7 years ago. The tapes are new. However, its not a problem when a note is being played (although I suspect that if there was less noise in the system, it would sound better).
The issue is, when no tape is being played, the machine makes a lot of noise and its all before the volume control.
The issue is, when no tape is being played, the machine makes a lot of noise and its all before the volume control.
When you get a chance try to post the transistor number and I will see if I can find a spec sheet on it.
Some a have high NF (noise factor)rating of like 6db to 10 db or more.
And some of todays transistors are tipicaly 3db or less.
I remember in my days of repair and tearing apart 8-track tape players(as I was obsessed with the idea of building my own super quality stereo system from scratch) that some head preamps were quite noisey and some were super quiet.
I noticed that the noisey ones used (A) type transistors and the less noisey ones used (B) type transistors and the preamp boards were all pretty much the same design style and construction using 2 transistors per channel.
I found this to be true when I had acsess to a ton of scematics and various stereo units that I worked on that came through the T.V. repair shop where I was working in the early 80's.
And then I discoverd opamps (in the late 70's).
FYI,For those whom just starting to follow this thread the mellotron's noise rating was -55db which was is a bit high even for the early to mid 70's. jer
Some a have high NF (noise factor)rating of like 6db to 10 db or more.
And some of todays transistors are tipicaly 3db or less.
I remember in my days of repair and tearing apart 8-track tape players(as I was obsessed with the idea of building my own super quality stereo system from scratch) that some head preamps were quite noisey and some were super quiet.
I noticed that the noisey ones used (A) type transistors and the less noisey ones used (B) type transistors and the preamp boards were all pretty much the same design style and construction using 2 transistors per channel.
I found this to be true when I had acsess to a ton of scematics and various stereo units that I worked on that came through the T.V. repair shop where I was working in the early 80's.
And then I discoverd opamps (in the late 70's).
FYI,For those whom just starting to follow this thread the mellotron's noise rating was -55db which was is a bit high even for the early to mid 70's. jer
Last edited:
OK- I finally had some time to open up the machine and document the schematic for the tape head equalizer. As you can see, it has some oddities- IMO it seems a bit daring to have such a large value collector resistor on the first stage, to begin with. The base of that transistor is biased in a rather odd way too.
I can't make out what the diode is- it has 3 bands on it and thus may well be oriented the other way, but given the circuit, I think this is correct and that it may be a zener diode. Also, the output device is labeled as 'MED41' which is not a valid type as best I can make out on Google. I have no idea why there are two coupling caps in series at the output. I triple-checked that: it really is set up that way.
IMO this thing is ripe for an update of this circuit! It looks like the original idea was to get as much gain from the first stage as possible; right now I am leaning towards an FET input stage and more current through the stage- I suspect the noise could be reduced considerably.
I can't make out what the diode is- it has 3 bands on it and thus may well be oriented the other way, but given the circuit, I think this is correct and that it may be a zener diode. Also, the output device is labeled as 'MED41' which is not a valid type as best I can make out on Google. I have no idea why there are two coupling caps in series at the output. I triple-checked that: it really is set up that way.
IMO this thing is ripe for an update of this circuit! It looks like the original idea was to get as much gain from the first stage as possible; right now I am leaning towards an FET input stage and more current through the stage- I suspect the noise could be reduced considerably.
Attachments
Best I could find is BC239, I remember my sunn concert lead amp had these in the preamp and it was very hissy when I turned it all the way up , which is were I kept it on a normal basis.
That was what inspired me to start designing my own preamps and distortion units in the late 70's using opamps that were quiet and did so with great sucsess.
After that I never turned back and learned transistor design except for learnig power amplifier circuitry.
I found some old data on those transistor and the claimed a typical NF of 2db to 4db and a maximum of 10db ,quite noisey.
However I am not sure what the HFE is, as some batches were as high as 900 according to the data sheet I found.
I will draw the schematic up in circuit maker and see what I can make of it .
The first stage appears to have a gain of 4700 by my rusty transistor calculations.
Good thing that there is only one preamp as I thought that there were 70 of them. jer
That was what inspired me to start designing my own preamps and distortion units in the late 70's using opamps that were quiet and did so with great sucsess.
After that I never turned back and learned transistor design except for learnig power amplifier circuitry.
I found some old data on those transistor and the claimed a typical NF of 2db to 4db and a maximum of 10db ,quite noisey.
However I am not sure what the HFE is, as some batches were as high as 900 according to the data sheet I found.
I will draw the schematic up in circuit maker and see what I can make of it .
The first stage appears to have a gain of 4700 by my rusty transistor calculations.
Good thing that there is only one preamp as I thought that there were 70 of them. jer
I've checked it several times. It is indeed correct. I plan to take voltage measurements soon- so the zener diode value will be obvious at that time. The lighting was poor enough that even with a flashlight I could not make out the color codes on the zener. Anytime I doubted the resistor values I measured them just to be sure.
Seems to me that the designer went for too much gain (and not enough current) in the first stage- I suspect that most of the noise originates in that circuit.
So there's really a 10uF and 0.47uF in series on the output? - why?.
Do 221K resistors (and other strange values) really exist?, I wouldn't have though so - or that people would use them?.
I would suggest making an opamp tape preamp, and trying that to see how it sounds - you could try something like this:
http://rbsfm.org/Downloads/Datasheet/LM/LM1897.PDF
The redundancy of output coupling caps is peculiar and I can think of no reason for it.
221K is the precision form of 220K- nothing unusual there.
I'm hoping for something a little simpler. The circuit board that this is on measures about 2 1/2" square. I really would only need one opamp, as there is a line section preamp that works fine. But it would be nice to not have to fabricate a new board. But I might be over-optimistic on that one.
221K is the precision form of 220K- nothing unusual there.
I'm hoping for something a little simpler. The circuit board that this is on measures about 2 1/2" square. I really would only need one opamp, as there is a line section preamp that works fine. But it would be nice to not have to fabricate a new board. But I might be over-optimistic on that one.
Last time I was at the studio I took voltage measurements of the tape head preamp. From them it was revealed that the output transistor is in emitter follower mode and is PNP. The voltage across the unknown diode shows that it is some form of a small signal diode- and apparently being used to lock the emitter voltage of the 2nd stage to a fixed value.
So now the circuit makes more sense- the cap and resistor bypassing it in the lower middle of the drawing is essentially a power supply leg for the final stage. The resulting voltage is used to bias the input stage. Its apparent that high gain at the input was a priority in the design, and I can see that even with the 1 volt value at its output, the input transistor would never be driven to cutoff. But it seems to me that with such operating parameters (0.0463mA of collector current!), that stage would also be excessively noisy.
I'm thinking of several changes. Lower noise transistors, more current on the first stage, and to deal with that, perhaps 2 or 3 diodes in series on the second stage, with more bypass capacitance for the bias network of the first stage. Finally, eliminating coupling cap at the output! Any thoughts?
So now the circuit makes more sense- the cap and resistor bypassing it in the lower middle of the drawing is essentially a power supply leg for the final stage. The resulting voltage is used to bias the input stage. Its apparent that high gain at the input was a priority in the design, and I can see that even with the 1 volt value at its output, the input transistor would never be driven to cutoff. But it seems to me that with such operating parameters (0.0463mA of collector current!), that stage would also be excessively noisy.
I'm thinking of several changes. Lower noise transistors, more current on the first stage, and to deal with that, perhaps 2 or 3 diodes in series on the second stage, with more bypass capacitance for the bias network of the first stage. Finally, eliminating coupling cap at the output! Any thoughts?
Attachments
Last edited:
I haven't yet tried simulating it yet as I got caught up in a nakamatchi tuner problem.
So, I will do so tonight.
I stumbled upon an intresting circuit in the 02/10/11 issue of electronic design on page 100 with a diode in a similar situation.
It may be there to perform a type of logarithmic function or some type of compression action, if you will.
I just downloaded a screen capturing program so now I can show my result from circuit maker easily after I learn how to use it.
The resistor and cap network your are refering to is a type of feedback network I have found this type of configuration described in one of my many reference text books and I will try to locate that as well.
Cheers. jer
So, I will do so tonight.
I stumbled upon an intresting circuit in the 02/10/11 issue of electronic design on page 100 with a diode in a similar situation.
It may be there to perform a type of logarithmic function or some type of compression action, if you will.
I just downloaded a screen capturing program so now I can show my result from circuit maker easily after I learn how to use it.
The resistor and cap network your are refering to is a type of feedback network I have found this type of configuration described in one of my many reference text books and I will try to locate that as well.
Cheers. jer
Last edited:
I have been doing some simulations and this is what I have found so far.
The 82k resistor and 2.2meg set the gain of the circuit.
As it is stock with a 2mv P-P signal input the output signal is 1.70v P-P.
if the 82k resistor is reduced then the gain of the circiut is reduced and if it is increased the gain is increased as follows per simulation.
input = 2mv p-p
ohms out Vp-p
0(shorted) .680v
39k .991v
41k 1.020v
82k 1.692v
100k 1.969v
220k 4.039v
470k 8.082v
1meg 14.239v
open 21.052v clipping
The waveforms do not show any clipping except when the feedback circuit is open.
The circuit exhibits a large subsonic peak around 10hz and this explains the extra capacitor on the output.
As I had figured was a last ditch effort to reduce the subsonic output from the noise that the circuit produces.
I once used this tecnique to dampin the subsonic effects from pushing on the strings of my guitar when I was designing very high gain low noise distortion boxes with opamps.
When you can please confrim all of the the values of the capacitors in the circuit as well as the resistance value of the output pot.
A change in the resistance value that loads the output capacitor changes the overall frequency response of the circuit simulation.
That way I can post an accurate freqency response simulation of the circuit.
Meanwhile I will try to figure out if I can get a simulated niose figure of the circuit.
Regards. jer
P.S moderator I posted this at 10:59 pm my time
The 82k resistor and 2.2meg set the gain of the circuit.
As it is stock with a 2mv P-P signal input the output signal is 1.70v P-P.
if the 82k resistor is reduced then the gain of the circiut is reduced and if it is increased the gain is increased as follows per simulation.
input = 2mv p-p
ohms out Vp-p
0(shorted) .680v
39k .991v
41k 1.020v
82k 1.692v
100k 1.969v
220k 4.039v
470k 8.082v
1meg 14.239v
open 21.052v clipping
The waveforms do not show any clipping except when the feedback circuit is open.
The circuit exhibits a large subsonic peak around 10hz and this explains the extra capacitor on the output.
As I had figured was a last ditch effort to reduce the subsonic output from the noise that the circuit produces.
I once used this tecnique to dampin the subsonic effects from pushing on the strings of my guitar when I was designing very high gain low noise distortion boxes with opamps.
When you can please confrim all of the the values of the capacitors in the circuit as well as the resistance value of the output pot.
A change in the resistance value that loads the output capacitor changes the overall frequency response of the circuit simulation.
That way I can post an accurate freqency response simulation of the circuit.
Meanwhile I will try to figure out if I can get a simulated niose figure of the circuit.
Regards. jer
P.S moderator I posted this at 10:59 pm my time
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.