Dr. König´s Märklin-Digital-Page

Booster with stabilised output voltage


From a functional point of view the Digital systems consist of a Control-Unit,that creates the digital information, and an amplifier / power-driver, commonly referred to as "Booster", that adds this information to the power-supply. Depending on the system, these devices may be physically separate, but with Marklin-Digital the new Control-Unit 6021 does also contain such a booster.
When the available power is no longer enough, for everyone the question pops up : " make or buy ? ".

With the booster there is only one answer for the hobbyist : "make", of course. Because the commercailly available boosters known to me have the big disadvantage that the output voltage is not stabilized - in spite of their fairly high price. The output voltage of the Marklin Control-Unit-Booster, and thereby the voltage of the digital signal on the tracks, fluctuates, depending on the load, between +/- 23 and +/- 15 V. It is obvious, that this causes very undesirable speed changes. Even the very good regulation of Marklin-6090-decoders cannot correct this. The lighting shows a disturbing flickering behavour because of this. An other reason for these voltage fluctuations is in the Marklin-transformers : they only supply fixed 16V AC, and are pretty weak, which means that with a serious load there is a considerable reduction in output voltage. It is of course very difficult to create a stabilised +/- 18V from that.

When building your own booster, one should be modest. 5A output current is enough in any case - may be even too much -, because with higher output rating, short-circuits may "weld" the weels to the rails, and cause other damage. I have had such experiences with my EDITS-booster, rated at 10A.


The home-made booster presented here, and described in detail in MIBA 5/99, fulfills these requirements of stabilised output voltage and maximum currents of 1.7A and 3A respectively. One can still use the standard Marklin-transformers. However, they have to be slightly modified. Alternatively, a better transformer than the Marklin one may be used.

I have to add a remark to above circuit-diagram : I did develop this circuit for the Marklin-system, but (originally) not for the Control-Unit. My intention was to build a small booster for LOK, that would be driven directly from the PC-serial-port, without control-unit or interface, quasi-digital, almost for free. Therefore I had to add a small "interface" for the connection to the Marklin-CPU, which enlarges the circuit a little bit.
In addition, I wanted to use the many 32VA Marklin-transformers, that I have as left-overs from beginners-sets. One has to use some tricks, when one wants to be able to use these transformers. These include "surgery" on the transformer, to be able to use the 24V AC which is available inside. One can, of course, also use a stable 18-22V transformer. The Marklin tranformers have the advantage, that they have an integrated overload- protection; and the high output impedance is compensated by the 24 V available.

With the addition for the 6021, the booster has shown to be able to operate with the Intellibox as well, without any change to that interface. Although it was not designed for this, it also amplifies the SELECTRIX digital signal in an acceptable way, at least as seen on the oscilloscope. This is due to the connection/disconnection of the power signal without the use of relais.
It is well-known, that the Control-Unit switches off the booster : the output becomes "floating" and there is no more voltage on the tracks. A well-designed booster must also perform this function; this is done by the components between JP2 and IC3. They prevent voltage on the tracks when the power-signal is absent. This is achieved at "electronic" speed,- very fast- and therefore also very usefull for the SELECTRIX. This system has the peculiarity, that it switches not only between positive and negative, but also, in between, to ground for about 10 microseconds. A relais is obviously not capable of switching at this speed, this can only be done electronically. I don't know if SELETRIX-decoders work well together with this circuit; since I don't have any, I couldn't test this. The circuit may have to be optimised for this application.


The connection of the booster to the CU is done on the CU-side to the plug dedicated to the booster; on the booster-side with the usual SUB-D-plug that we know from the PC-serial-port. The pin-connections on the CU-plug are (looking to the backside, starting from the left ):
pin 1 short-circuit
pin 2 ground
pin 3 open
pin 4 power
pin 5 data

The SUB-D-plug should be wired as follows :
pin 1 short-circuit
pin 2 data
pin 5 ground
pin 8 power

The cables that lead to the tracks are connected with two outlets for the standard Marklin plugs ( 2.6 mm miniature plugs) The cable for the mains is lead through a hole and safeguarded against mechanical pull with a wire-clamp. The picture shows the back of a prototype connected to a 32VA transformer that has been modified. Prototyps.


As mentioned, the 16V AC of the 32 VA transformer is not sufficient; we have to modify it.

When looking at the sides, one can observe two small plasic plugs. These can be removed by brute force, by drilling trough, but then one has the problem that is no longer possible to close the box again in a way that is child-proof. It is more elegant to drill small 1.5 mm holes in the plugs. If one is lucky, the pins will attach to the drill, and can be pulled out. If not, one can take a fitting screw and screw it into the hole. When it is securely fastened, one can "pull the plug". Afterwards, it is advisable to make an M2 or M2.5 thread in the plug, so that it becomes easier in the future to remove these plugs, by simply screwing in a fitting bolt, and pulling out the plug.
However, in newer trafos it has become increasingly difficult to remove the plugs in this way. In that case, one has no other choice but to completely remove these plugs by drilling. One should take care, only to remove that part of the plug, that holds the top and bottom parts of the box together. In this way, one can later use screws to close the box again. After removal of the four plugs one can remove the cover; one has to be carefull with the wire between the two parts. The 24 V connection is now visible on the left-hand side; it is the bottom one of the two copper areas for the contact of the direction-change switch. A cable with appropriate cross-section has to be soldered to this area, and to one of the two brown ground-connections, which before has been isolated from the overal ground. The picture to the right shows the open transformer box after above changes have been made; the connection of the 24V is on the bottom side of the PCB and not visible in the picture.When the box is closed, please make sure that the cables are laid correctly, and not squeezed anywhere. After repositioning of the plugs ( with M2 threads inside ) one should distinguish the new 24V output plug from the others by giving it a new, different color.

It is understandable that Marklin does not recommend this procedure; after this you can't any more go back to your dealer with any kind of damage, since he will claim that it is caused by your actions - it is not worth the effort, with the (low) price of these transformers. I have modified several of these transformers, and operated them over longer periods of time without any excessive heating - which is no surprise, since Marklin must have taken care of the possibility that children turn the knob "left" (=24V) for longer periods of time. (I did in my youth)

The two single-phase rectifiers allow a safe parallel use of two transformers. It is well known that the parallel connection of the secondary windings of two transformers, can cause 220 V to appear on one mains plug, when it is - by -accident, pulled out of the wall-outlet. The shown circuit avoids this problem; it is however recommended that both mains cables from the two transformers be soldered together and connected to mains via one, shared, plug. Attention should be paid to the polarity (phase ) of the transformers. This can be done by connecting both brown outlets, and when one then briefly connects the yellow ones, no short-circuit should be observed. Instead, one can also connect both newly created 24V connections by a regular 220V bulb : when the polarities are the same, nothing will happen. In the other case, it will light-up dimly.
It has been observed, that, in particular when two transformers are used to power the booster, that a short circuit on the tracks, when it is more than 2-3 m away from the power-supply point of the tracks, is not recognised by the 6021. This can be explained by the relatively high series resistance of the tracks: there is not enough current flowing, to achieve the necessary voltage drop in the booster. One should - not only for this reason - make sure that power connections to the tracks are made roughly every 100 cm, and use cables with appropriate cross-section. The voltage loss at the contact-resistances between the tracks cannot be negelected. In addition, the Zener-diode D16 can be changed to a 20V type.

Finally one remark : bulding and using this booster, as well modifying and using the transformer is at your own risk.

Parts & Materials

As mentioned, there is an extensive manual in MIBA 5/99 and in German (as updated and debugged version, too) when clicking below at the German symbol. You can also find the PCB layout in there; also here for download. It can also be obtained, as usual, from Mario Binder; ifso required, he can also supply the special plugs for the 6021 connector.


This translation was made by Ir. Roelof Salters from The Netherlands. Many thanks to his grateful work. Some mistakes you probably find are my work as result of trying to correct some misunderstandings and add some additions.

Homepage | Deutsch

© 1999-2000 by Dr. M. Michael König | Antoniter-Weg 11 | 65843 Sulzbach/Ts. | Contact | Stand: 19.10.2000
© 1999 by Ir. Roelof Salters | c/o Philips Research Laboratories | Building WAY4 001, Prof. Holstlaan 4 | 5656 AA Eindhoven (The Netherlands) | salters@natlab.research.philips.com

Revisit this page