Inverted Variations on a clone?

Part 2: Proto Construction

Last modified: 30-May-2004


This page describes the construction process of the Cyclone project. Since this is a team-effort of Jerry and myself, it may take some time before the project is finished.

Anyway, we'll make two complete amps, one for each of us, and it may even be that these two amps differ in components used, but this is not our current intention.

Bodyworks of the Amp

Jerry thinks that the following layout of the components in the small boxes should be the way to go:

The red square is the rotary attenuator switch for the volume control (I'm currently having a difficult time to find one that fits). The two blue rectangles are the two 1000uF caps. At this moment there is an on-off switch foreseen....

The prototype resulting from the schematics above including connectors on the backplate looked very satisfactory and leaves enough room for the internal (9) components and the attenuator switch.

Above you see the first prototype of the housing including connectors. The power cable will be fixed at the amplifier side and probably a little clamp will prevent damage to the wire when people pull on it. As you can see we will use standard binding posts and input connectors, the original Gaincard uses chassis connectors that are optimized for their own speaker wire and interconnects.

Continued (Oct. 2003)

Some months after the first prototype of the amp was made, we decided to pick up the project again in Oct. 03 (last major work was done April 03). Jerry finished the housing of the amp, and the "final" box now looks like this:


It shows quite well the final layout of the inside of the clone, and it is shown how the chip will be mounted on a sort of table. The power switch on the front will probably not make it. I do not see the use of it anyway apart from making this clone look closer to the original. And I think we made a better version looking at the backside of the amp.

And after a days work the proto looks like this. Power connectors are placed, and two attenuators (12-step). This was also the moment for deciding on the type of capacitors to use. The current table was just 5 to 6 mm too large to accomodate Black Gate capacitors. And since we would at least have one amp with Black Gate capacitors, Jerry would modify the "table" where the chip is mounted on to make room for the larger caps.

With that idea in mind, I started to build the prototype with Black Gate capacitors, which are so much larger than the standard caps that when these fit every other cap will fit as well. Hmm, I discovered that the "table" was still to large, because when the connecting wires of the caps are bent they add another 2 mm to the lenght of the caps, and that as a result the caps would not fit between the table and the opposite wall of the cabinet. It cost me two hours to remove 2 mm of the table because I do not have the nice tools Jerry has.

And a little later (well, lets call it a day, time flies when you're having fun) the picture below shows the "finished" amp. Still, this is the prototype with open tubes for both channels. It should be easy to take the guts out and mount them in a the final tubes which a closed type.

Some notes to go with the last photo:

And since you ask: Yes they work and Yes I like the sound already. I have to wait until the Black Gates have had time to break in, but normally after a day or two they reach a level which is close enough to give an idea of their final sound quality.

Final box design (empty model)

I'm now waiting for Jerry to get the final parts ready after which I'll build two amps. The following things still need work:

Final Steps

On page 3 you find our listening impressions with the prototype. they were very positive and therefore Jerry and I decided to build two production clones which are identical to the prototype but with better bodywork (and hood closed).

Anyway,here is another photo of the volume control section, with prototype LED in the front. As you can see, I only connected one LED in the knob of the right channel, and as I write this down I'm listening to the clone, but I'm unable to hear any influence of the LED on the sound quality (connected the LED to V+ and ground with a 16k resistor in series). It's not the LED itself that's giving me headaches, its the wiring inside the knob which has to be flexible enough to stand a 320 degress turn of the knob without breaking or shorting (duh).


The attenuator is a cheap but reliable Lorlin type (for the moment) and I found break-before-make types only, therefore there will be an additional 47k resistor between the switch and ground. I'm looking for a make-before break type, to replace this solution later.






1 (low) GND 1 -49,24 82
2 1 2 -43.24 82
3 2 3 -37.62


4 3 4 -31.44 330
5 4 5 -25.29 680
6 5 6 -19.87 1,200
7 6 7 -15.04 2,000
8 7 8 -11.56 2,400
9 8 9 -8.58 3,000
10 9 10 -5.70 3,900
11 0 11 -2.87 4,700

12 (hi)

11 12 0.00 5,200

However, the above resistor values are quite valid for non-inverted clone designs, as the input volume control is not a part of the feedback loop. However, for inverted configurations the attenuator is part of the feedback loop. There are several potential issues to consider:

Therefore, it is important to take the total feedback loop into consideration when calculating the gain. This will result in different values for the feedback loop.

See the chapter on Attenuators for Inverted Amps, in the Background section.

With an effective input impedance of 17k, this design is optimized for sources with a low output Z. Unfortunately tube phono amps do not fall into this catagory. When redesigning the volume control section we might take into account.

Bodyworks of the Power Supply

The power supply will be built the same way as the original. Moreover, Jerry would like to make a similar tube-shaped housing for the power supply with the same sort of massive aluminum front- and backplate.

The power supply will be built as follows:

The black square is the IEC 320 chassis connector and the red will be the DC power connector for the amplifier. The print will contain the 8 diodes (4 per channel) and 4 small capacitors (100uF).

The small red bult on the front-plate is the LED (on the original Power Humpty only on recent models (?)) Below you find the picture of the power supply in the prototype stage:

The following photo shows the prototyping of the Power Supply: We need to fix the bridge/rectifier section on a board in the big cylinder. I do not yet know whether I have to take further precautions for the board, but given its limited weight I presume that using the central hole is enough to keep it in position.

As the tube has enough space to accomodate another transformer, I will populate only half the board leaving room for further experiments with double transformers (true monoblock operation).

Just a few photos of the finished prototype. The two "production" models will be made black and will most likely have just one output (we'll use dual mono with two separate power supplies for this design). Also we will make use of a DIN connector on the power supply as well (although it is understood that his may introduce another connection, it will allow more flexibility). on the other hand, inside the tube it will mean a much simpler connection to the power supply board and thus the nett effect will be minimal.

The power LED in the power supply may disappear in the final version too. After all we have small low power LEDs in the knobs of the amp so there will be a visible indicator of power-on. The most difficult part will probably be to get 2 additional IEC power entry modules, since the ones we use here with a IEC320 power entry, fuse and switch in one module are difficult to get in NL.

Umbillical Cord (Power Supply)

For the connection between the amp and the Power Supply I made a powercord along the same lines of the QuaDraad mains power cable. Getting the DIN connector on the wire without risking short-circuits later is not easy. Use crimp sleeve for insulation especially when soldering two wires to one pin (shown on the black wire).

I used two black wires for V+ and two white wires for V-. The red wire was used for ground. The red wire straight and the two white and black wires braided as in Quadraad I made two cords of 75cm nett length.
Initially I connected the amp to the Power Supply of GeenKloon using these cords until its own Power Supply was finished.

Still we have to decide whether the power cable needs protective sleeve or so. Shielding may not be necessary because of the way the cable is constructed.

Electronic Parts

I'm planning to use good-but-not-exotic components in this amp. One of the major changes apart from the inverted mode will be the selection of the LM3875. Unlike the GeenKloon I will make use of the non-isolated package for the chip. The Geenkloon uses the LM3875TF version which is isolated plastic and does not dissipate power as well as the non-isolated version. As a result it's output power is lower also.

The following parts will be used for the Cyclone:

Amp National LM3875 chip amp LM3875T
C1 Input (DC) cap, Black Gate N type 4.7uF
C2 n.c.  
C3 Elco 1000uF, Philips BC, or Black Gate 50V/1000u as an alternative 1000uF
C4 Elco 1000uF, Black Gate 50V 1000uF
R1 R2/R1 determines Gain, Holco or Dale 10K
R2 Feedback: Caddock, Dale or Holco 200K
R3 Minimalises DC offset, same type as R1 10K
R4 Input Impedance, determined by the stepped attenuator value. 22K

Power Supply

T1 Power transformer Amplimo 68015 2*22V/5.1A
N1 IEC320 net entry, with fuse and switch  
F1 Fuse 1A, for IEC320  

Diodes, Fast Recovery type 860

Cd1-Cd4 Capacitor 0.1uF
C1, C2 Small rectification PS 100uF
S1 power Switch  
L1 Red Power Indicator LED (5mm) Blue
R_l1 Resistor for LED 5K5


Once the amp is finished we will publish listening impressions. Initital listening impression is found on page 4, based upon this 2 hours session jerry and I gave the final stage a "go" so we're starting the after-proto phase now. We need to finish the housing either in champaign or black and with these I will build 2 "production" quality amps.

<< Page 1: Intro and Design


©, Feb 2003