Introduction:

A fellow enthusiast used my Class 50 rear lights disable transistor circuit, when fitting a Bif LokSound V4 & speaker system to his locomotive.  However, when the rear lights were supposed to be disabled, a slight glow was still clearly visible. After a few ideas had failed to resolve this problem remotely..... I decided the best way to investigate the issue and find a fix was to make a similar conversion to my own 50.  That's where this mini project started:

Nothing very complicated!  A PNP transistor is inserted in the positive supply to both rear light LEDs. When the PNP transistor is biased "on" via a decoder function output, the rear lights behave normally, with their negative supply provided from the decoder directional (white and yellow) function outputs.

The problem and the fix:

The original decoder in my class 50 was replaced with a LokSound V4 (recently freed up by a change to my Class 31). When running on a short test track, the problem was visible. With the rear lights theoretically disabled by the removal of their positive supply by the transistor, a very faint glow was still discernable. Curiously on Marcel's locomotive, this glow got brighter when the loco moved (although I could not reproduce that extra effect on mine). 

DC checks with a voltmeter & some guesswork:

All appeared normal, with no suggestion of a voltage across the LEDs to enable them to turn on.  However, that glow was not imagination!

Now without an oscilloscope, I can't actually look at the real time dynamic voltage. My meter just gives me an average DC reading. So the next bit is speculation.... If a sequence of very short negative going pulses is present at the function output driving the transistor base, this could produce a result like this and would be invisible to my meter. If present, these could be reduced in amplitude by adding a suitable capacitor between the transistor base and emitter, forming a simple low pass filter with the base resistor.

A check in the spares box revealed a 10N capacitor, which felt the right kind of value. If this made a difference to the intensity, it would be a good indication that the speculated mechanism for the problem was correct and the value could be adjusted as required.

However, as soon as the capacitor was added there was no further sign of the problem.   A quick email to Marcelo in Brazil resulted in confirmation that the same fix had also cured the issue on his 50. (even when it moved.)  

Bif's new 50 sound project includes synchronised outputs on Aux3 to drive a LED flash every time his detonator bang sounds. So the first mod. was to add a tiny chip LED just above and beside to the rail contact point of the front wheel of the leading bogie.  

The only speaker option I could come up with involves removal of the fan assembly. I've mounted a bass reflex speaker (with 20x40mm drive unit) facing downwards in this location.

I've also added a modern casually dressed scale driver, in line with rail-tour excursion operation, which is the 50's new role . (Replacement of the cab assembly was a bit of a nightmare!)

My LokSound has now been sent back to Bif for a re-blow with his new Class 50 software.

Swiftly returned with the new software on board..........

Before running, some minor adjustments were made to the drive control optimisation CVs (55) and the master volume (CV63) was reduced from a speaker shattering 192 maximum to circa 60.

The loco then ran smoothly, with Bif's sound system responding well.

Bif's synchronised detonator flash LED drive on Aux 3 is spot on!  Providing a brief flash to correspond with the detonator bangs across the entire speed range. (The bangs get closer together as the speed increases, simulating a fixed distance along the track between the detonator packs.)

However, my speaker choice requires a rethink as the unit I have fitted sounds decidedly lacking in bass. (A 1 Watt Zimo bass reflex design using a 20x40 mm drive unit.) The sound seems very "toppy" with over emphasised higher frequency content.  In an attempt to partially compensate, I increased the volume of the Hoover fan sound on F12 significantly. I've ordered another unit from a different manufacturer and I will see if it behaves better...... or if I can find an improved way to mount it.

I also increased the volume of the dispatcher whistle. However, the detonator bang volume, which I would have liked to increase, was already at maximum.

Further experiments with the Zimo 1W reflex speaker: I've discovered that the output treble content can be reduced by two steps that seem completely counter-intuitive!

I've ended up by covering the reflex port (which was actually boosting mid to treble frequencies rather than bass) and by remounting the speaker so that it faces upwards. (The aperture in the chassis block that it faced when in the original downward facing position, was creating some odd resonances and accentuating the treble).  Its quite a bit better than before, but this has more to do with reducing high treble white noise effects than boosting any bass content.

I'm going to try some alternative speaker solutions:

1) I've sourced one of DC Kits etc EM2 speakers from Charlie at the Peterborough show..... This is reputed to have an exceptional bass response, but it would require surgery to the chassis block to fit inside. (So I'll only use it as a last resort.)

2) I have a 4 ohm bass reflex speaker, which looks identical to the "Zimo" 8 ohm unit. I've tried it outside of the Class 50 chassis, where the reflex port is definitely assisting the bass output... I also discovered that somewhat lower frequency bass can be accentuated by adding a short pipe to the reflex port.

3) If the 4 ohm bass reflex is no better than the original 8 ohm unit, I'll try a trusty ESU 20x40mm 50334 speaker in a custom enclosure, based on suggestions from Mike B.

4) Marcelo from Brazil has used a more sophisticated speaker arrangement on his 50...... I'm waiting to see and hear his video to see if gives any clues for further enhancements.

1) EM2:  Its a curious looking device!  Two drive units in a large sealed enclosure. (Impedance measures 4 ohms so probably a pair of 8 ohm devices in parallel.) Once the two substantial bolt down  flanges are removed, size is approx 65x29x12mm. The bass output is impressive. I taped it to the chassis face up and the sound quality improved a lot when a cupped hand was moved over the top. There was a definite bass thump that transmitted to the baseboard via the wheels (with or with out the cupped hand.)  To fit it I would need to hacksaw the chassis block sides (unfortunately, no access to milling technology) but I would also have to rebuild the opening side louvres as fixed flush parts to get the width of the speaker past these during body fitting. So I'll try other solutions first.

For my own future reference, the Hoover fan on key 12 is also sound slot 12. CV32 to 1 then CV347 sets the volume ..... I used a value of 36 on this speaker.

2) 4 ohm bass reflex:  No where near the bass clout of the EM2 but face down with the reflex port nearest the loco centre, I think it will do, but CV347 will have to go up a bit..... volume dropped & CV347 up to 65-70!

If I can find a reference Class 50 to listen to, I may come back, but I'll run with this arrangement for the moment.

After a more critical look, I think it is just possible to squeeze the EM2 in at the opposite end to the fan louvres, by removing the original main board completely and re-mounting the LokSound decoder at the fan end. Raised parts of the chassis block sides definitely have to be removed and despite some fairly aggressive chamfering of the rear corners of the EM2 box, I can't quite get it high enough into the roof to avoid also having to file a thin chunk of metal off the remaining side walls over the entire length of the EM2. I'm planning to mount the speaker facing downwards, glued to the remaining edge of the chassis block. I'll also have to file out recesses to allow the drive unit cones to move over their full travel.  I may also have to remove a small part of the EM2 box end to avoid hitting the motor fly-wheel, in which case, I'll seal the hole with rigid material and try to make up the volume higher in the roof section.

Preparation:

I've removed the main board, the decoder and the speaker, to give unrestricted access to the top of the chassis block.

I've also removed the bogie from the end I plan to attack with a hacksaw and some files.

The final process before metal bashing is to protect the remaining moving parts (motor, flywheels drive shaft and the far end bogie) from the metal swarf I am about to produce. (Polythene bag over the far end bogie etc and electrical tape, paper and blu tack over the drive shaft, flywheel and motor shaft / bearing at the cut end.)

I've also made a pair of plywood spacers with small slots cut in them to accommodate the upper body clips that protrude from the side of the chassis block, when the assembly is clamped in my vice.

Showing the moving cones on the side of the EM2 that will be in contact with the chassis block.

Below is a rear view of the EM2 showing the chamfered edges that touch the 50s curved roof underside.

Well, that went well (I think). The speaker now fits inverted over what's left of the chassis block and I can mount it just forward of the original cab door bulkhead which has now been removed. This takes it well clear of the motor flywheel and crucially into the region where I have been able to file the inside edges of the chassis bloc walls thin enough to avoid any contact with the moving drive unit cones. The circular cone is completely clear, however if the rectangular diaphragm has a lot of movement, it may still be restricted by the sidewall.... we'll see!

 I also added new longer wires to the track contacts on the bogie and refitted it. Then with just the motor, track contacts and speaker connected up to the decoder, I did a brief sound test to see if there was any sign of cone bumping on the chassis.

Seems OK, no extraneous noises so far.

Next a check on the lighting arrangements using an ohm meter to re-identify which wires are connected to which LEDs. I've used a different circuit option that removes any need for the original PNP transistor (circuit shown below). Directional lighting is now controlled by key zero for the driver's end and key 20 for the other end. The number two end wiring now runs over the top of the speaker through the small central gap between the back of the speaker and the roof underside.

The revised circuit diagram

Function mapping: Key zero = end 1 lights (Fwd R/L, Rev Aux2); Key 20 = end 2 lights (Fwd Aux1, Rev H/L)

Bif's on-board software drives Aux3 to produce the Detonator flash LED synchronised to the detonator bangs.

I've added 3000uF of stay alive capacitors (3 x 1000uF 25V capacitors connected in parallel. These are coupled to the decoder ground and common positive via a diode and a parallel pair of 1k resistors.

Time to refit the upper body shell.

The photo of the real Class 50 was taken on the MNR at Dereham in Feb 2010.