Roland RE-201 Space Echo - analysis

Discussion regarding early stompbox technology: 1960-1975 Please keep discussion focused and contribute what info you have...
Post Reply
User avatar
Barbouze
Information
Posts: 7
Joined: 10 May 2018, 10:30
Has thanked: 9 times
Been thanked: 1 time

Post by Barbouze »

Hi!

First post here so I'm introducing myself :D .
I'm not a guitarist (booo!) but I'm interested in guitar FXs, sound design and audio DSP. Lately I've been using a modeling tool to generate audio plugins from schematics (like a simplified spice simulation in real time) and after successfully creating a few simple circuits I'm trying a complex one, namely the amplification stages of the Roland RE-201.
Although I've been refreshing my (basic) electronic knowledge, I'm having troubles understanding some design choices and schematic annotations.

For the rest of this post/topic we can refer to the RE-101/201 service manual available here. A bloc diagram can be found page 4 and the detailed circuit page 6.

For the now I'm stuck with the first mic amp in the echo path, the one with 3 transistors.
After some research and a spice simulation I have some questions:
- reading the schematic, Q1 is biased @ 5V and its DC operating point is indicated @ 12.5V but a LTspice simulation gives me respectively 5.57V and 13.38V. The other given values (14V and 7V) are correctly simulated. What should I trust?
- there are two dashed boxes indicating respectively 9mV and 390mV. What would those values represent? Would it be the expected RMS value of a test signal?
- How is the feedback (R33, C25, R28) exactly working? This design exceeds my knowledge on transistors... Can it be modified to be connected to Q1 base instead of its emitter?
- what are the role of C27, C28 and C29?

Thanks for any tip! :)

User avatar
deltafred
Opamp Operator
Information
Posts: 1654
Joined: 06 Apr 2010, 16:16
Location: England
Has thanked: 813 times
Been thanked: 306 times

Post by deltafred »

Welcome to FSB.

2 things come to mind.

1. The Spice simulation is assuming ideal resistors (with values exactly as marked) whereas the schematic is using voltages measured from an actual circuit where the resistors have a certain tolerance. These could be 5%, 10%, or even 20% so the biasing of the first transistor will only be as accurate as the resistors in the potential divider and the dropper resistors that reduce the 17v supply down to feed it.

2. If you are not simulating the rest of the schematic fed by those dropper resistors they will not have as much current passing through them so will no drop as much voltage which could be another reason why your voltages are higher.
Politics is the art of so plucking the goose as to obtain the most feathers with the least squawking. - R.G. 2011
Jeez, she's an ugly bastard, she makes my socks hurt. I hope it's no ones missus here. - Ice-9 2012

User avatar
Barbouze
Information
Posts: 7
Joined: 10 May 2018, 10:30
Has thanked: 9 times
Been thanked: 1 time

Post by Barbouze »

Thanks!
For the now I've been simulating the circuit attached in this post, replacing the transistors with two 2N5089 and one 2N5087 instead of the original ones.
You're right about tolerances, this can lead to dramatical changes and given that this is an official manual, I'll stick to the given voltages instead of component values.

This leads to another problem, related to my fourth question.
As I'm using a simulator, I can craft any component I need so I don't have to use voltage dividers or shared voltage supply.
I would like to be able to power each transistor individually with its own perfectly crafted supply but as I don't understand what are C27 and C29 needed for, I'm not sure I can do this without affecting the signal/sound character. Is it doable?
Attachments
RE-201.png

User avatar
deltafred
Opamp Operator
Information
Posts: 1654
Joined: 06 Apr 2010, 16:16
Location: England
Has thanked: 813 times
Been thanked: 306 times

Post by deltafred »

C29, and C23, are power supply decoupling capacitors. They filter out power supply noise. This becomes more important with small signals and high gain.

C27 bypasses R29 (effectively shorts it out to AC) so the preamp will behave differently to AC than it does to DC.
Politics is the art of so plucking the goose as to obtain the most feathers with the least squawking. - R.G. 2011
Jeez, she's an ugly bastard, she makes my socks hurt. I hope it's no ones missus here. - Ice-9 2012

User avatar
Barbouze
Information
Posts: 7
Joined: 10 May 2018, 10:30
Has thanked: 9 times
Been thanked: 1 time

Post by Barbouze »

deltafred wrote:C29, and C23, are power supply decoupling capacitors. They filter out power supply noise.
So in a simulation they are not needed :D
deltafred wrote:C27 bypasses R29 (effectively shorts it out to AC) so the preamp will behave differently to AC than it does to DC.
I suppose this means that basses are treated differently than highs, as the lower the frequency is the closer it is to DC, am I correct?

User avatar
Manfred
Tube Twister
Information
Posts: 1936
Joined: 04 Apr 2009, 23:42
Has thanked: 1668 times
Been thanked: 1344 times

Post by Manfred »

The circuit is a direct coupled amplifier which provides a flat frequency response down to very low frequencies.
The first stage is a boostrap circuit which has an high input resistance.
The resistors R28 and R33 forms the negative feedback loop with an feedback factor of 0.0023
A rough calculation gives a gain of about 50 for the stages with Q1 and Q2.
The gain of the stage with Q3 has nearly unity gain and so the open loop gain is about 2500.
That results in a closed loop gain of about 370.
The capacitors C24 and C26 avoid oscillations.
C28 in parallel with R31 determines the upper cut-off frequency of about 30kHz.

User avatar
deltafred
Opamp Operator
Information
Posts: 1654
Joined: 06 Apr 2010, 16:16
Location: England
Has thanked: 813 times
Been thanked: 306 times

Post by deltafred »

Barbouze wrote:So in a simulation they are not needed
Correct.
Barbouze wrote: suppose this means that basses are treated differently than highs, as the lower the frequency is the closer it is to DC, am I correct?
You have to calculate the corner frequence (3db point) of the filter to know where the change occurs but just looking at the schematic with a large value capacitor and a resistor in the k ohms range it will be well below the range of guitar frequencies, almost DC.

Rather than break out my old calculator and I used http://www.muzique.com/schem/filter.htm which gave a corner frequency of 1.1Hz.
Politics is the art of so plucking the goose as to obtain the most feathers with the least squawking. - R.G. 2011
Jeez, she's an ugly bastard, she makes my socks hurt. I hope it's no ones missus here. - Ice-9 2012

User avatar
Barbouze
Information
Posts: 7
Joined: 10 May 2018, 10:30
Has thanked: 9 times
Been thanked: 1 time

Post by Barbouze »

There are a lot of informations in your two posts that I'll have to perfectly understand but this has helped me a lot! :thumbsup
Manfred wrote:The capacitors C24 and C26 avoid oscillations.
Are these oscillations related to the Miller effect?

User avatar
modman
a d m i n
Information
Posts: 4888
Joined: 19 Jun 2007, 16:57
Has thanked: 4392 times
Been thanked: 2130 times

Post by modman »

I retitled the thread. You can ask any question you like, but please use a descriptive title with at least the unit name.
In that way, other interested in this will be able to find it.
Please, support freestompboxes.org on Patreon for just 1 pcb per year! Or donate directly through PayPal

User avatar
Manfred
Tube Twister
Information
Posts: 1936
Joined: 04 Apr 2009, 23:42
Has thanked: 1668 times
Been thanked: 1344 times

Post by Manfred »

Barbouze wrote:There are a lot of informations in your two posts that I'll have to perfectly understand but this has helped me a lot! :thumbsup
Manfred wrote:The capacitors C24 and C26 avoid oscillations.
Are these oscillations related to the Miller effect?
The conditions those cause to oscillations are a positive feedbackloop to the input of the stage together with a sufficiently high gain.
Parasitic capacitances can form a positive feedbackloop.
The Miller capacitane forms together with the input resistance a low-pass filter that attenuates frequencies above the filter's cutoff
and thus prevented the oscillation conditions.
The input resistor could be the generator resistance or the output resistance of the previous stage.
What remains to be noted is that capcitors forms together with the input resistance a negative voltage feedback too.
The higher the frequency the lower the stage gain, but this will done in the radio frequencies range.

User avatar
Barbouze
Information
Posts: 7
Joined: 10 May 2018, 10:30
Has thanked: 9 times
Been thanked: 1 time

Post by Barbouze »

I spent some more time figuring how each part is doing its job and understood in the process that the numbers in dashed boxes were RMS values of the 1kHz -50dBm test signal.
This also confirms your statement:
Manfred wrote:A rough calculation gives a gain of about 50 for the stages with Q1 and Q2.
Correct! from the simulation we are at around 43 for each stage.

An interesting thing is that the amplifier used after the read heads and called pre-amp in the bloc diagram is almost the same as the mic amp with three minor differences:
- C26 has been removed.
- R33 has been replaced by a capacitor (C59) with an unclear value of 0.0068. Is it 6.8nF? Why this change in the feedback path?
- there is a capacitor (C54) linking the voltage divider biasing Q1 and its emitter. Why is it placed this way?

User avatar
Manfred
Tube Twister
Information
Posts: 1936
Joined: 04 Apr 2009, 23:42
Has thanked: 1668 times
Been thanked: 1344 times

Post by Manfred »

there is a capacitor (C54) linking the voltage divider biasing Q1 and its emitter.
Sorry I was wrong, this circuit with the link by C54 is the Bootstrap-circuit not the ciruit about Q1,
I probably mixed-up the ciruits on my explanation.
I have to find out, why R24 is built in, I did not ever see in this form of circuit.
R33 has been replaced by a capacitor (C59) with an unclear value of 0.0068. Is it 6.8nF? Why this change in the feedback path?
The NFB was frequency-independent using R33, after having replaced by C59 it is frequency-independent now.
The total gain must be high now for the audio frequency range, the NFB attenuate higher frequencies above,
thus C28 can omitted.

User avatar
Manfred
Tube Twister
Information
Posts: 1936
Joined: 04 Apr 2009, 23:42
Has thanked: 1668 times
Been thanked: 1344 times

Post by Manfred »

I have to find out, why R24 is built in, I did not ever see in this form of circuit.
I made a trial of the first stage using a BC548B resistor.
R24 does not have a significant effect on the biasing and the stage signal gain,
but it increases the signal input resistance form about 30 to about 55 Kiloohms.
You see, the subject had me hooked. :)

User avatar
Barbouze
Information
Posts: 7
Joined: 10 May 2018, 10:30
Has thanked: 9 times
Been thanked: 1 time

Post by Barbouze »

Manfred wrote: The NFB was frequency-independent using R33, after having replaced by C59 it is frequency-independent now.
The total gain must be high now for the audio frequency range, the NFB attenuate higher frequencies above,
thus C28 can omitted.
Probably a typo but I suppose you mean that C59 makes the circuit frequency-dependent?
The pre-amp is fed with a signal coming from a magnetic tape that is biased @60kHz, I suppose the NFB with C59 takes care of filtering this signal? The problem is that in this case if we are using 0.0068uF the lowpass filter
has a corner frequency of 86.7kHz (using @deltafred online calculator) and 86.7Hz using 0.0068mF. I would be tempted to say that C59 is 0.0068mF because using magnetic tape for recording exhibits a -3dB high pass filtering behavior so this would get us rid of the biasing signal and restore a near flat frequency response but I'm unsure about that :hmmm:

User avatar
Manfred
Tube Twister
Information
Posts: 1936
Joined: 04 Apr 2009, 23:42
Has thanked: 1668 times
Been thanked: 1344 times

Post by Manfred »

obably a typo but I suppose you mean that C59 makes the circuit frequency-dependent?
The pre-amp is fed with a signal coming from a magnetic tape that is biased @60kHz, I suppose the NFB with C59 takes care of filtering this signal? The problem is that in this case if we are using 0.0068uF the lowpass filter
has a corner frequency of 86.7kHz (using @deltafred online calculator) and 86.7Hz using 0.0068mF. I would be tempted to say that C59 is 0.0068mF because using magnetic tape for recording exhibits a -3dB high pass filtering behavior so this would get us rid of the biasing signal and restore a near flat frequency response but I'm unsure about that
Yes it was my typo, you are right, thanks for the hint.
The value of 0.0068uF is the right one, thus the corner frequency is 86.7kHz.

User avatar
Barbouze
Information
Posts: 7
Joined: 10 May 2018, 10:30
Has thanked: 9 times
Been thanked: 1 time

Post by Barbouze »

Back at it, I managed to get a better grasp of what is going on and broken down the power supply for each key component/bloc.
I also found out that BC550C and BC560C are the recommended replacement transistors for the original 2SC-1000 GR and 2SA493 GR. I needed only minor tweaking in the test signal to get expected gains and the result is in the attachment that contains two LTspice schematics for simulation for those interested.

In the simulator I'm trying to use, transistors are modeled using only:
- Vbc, base–collector voltage
- Vbe, base–emitter voltage
- Is, reverse saturation current
- Bf, forward common emitter current gain
- Br, reverse common emitter current gain
- Vt, thermal voltage.

Also, the internal sample rate won't exceed 96 KHz, leaving aside anything above 48 KHz.
This is far more simple than any SPICE simulation and of course real life circuits and I'm sure the circuit can be trimmed down. Any tips on what could be changed?

In the mean time, I'll study bootstrapping in amplifier circuits.
Attachments
amp tests.zip
LTspice schemas
(2.64 KiB) Downloaded 122 times

Post Reply