Low Cost DC and DCC Block Detectors

I tried using the Twin-T some 30 years ago and ran into problems. Back then I resolved none of them. So to start this project I tried to find as much info as I could on the Twin-T and its use.

Some History

In the June-August 1958 issues of Model Railroader magazine, Lin Westcott presented the design for a new, train block detector based on relatively new electronic components called… transistors. He called it the Twin-T because it was based on a pair of transistors (later including a third) for the basic circuit. Over the years it was updated to replace the ancient transistor antiques that Westcott used with some added modifications. Bruce Chubb, who was an early adopter of the Twin-T (MR Oct 1970 and MR May 1972), and later abandoned it for his own optimized detector design. In the August 1980 MR, Don W. Hansen wrote “The Twin-T Revisited” which offered several variations with updated transistor components. My revision is based on one of Don’s basic designs.

The recent request for help noted that the Twin-T circuit lit its indicator when the block was empty (original variation), and that needed to be reversed. Further, this need to operate multiple remote panel, block occupancy LEDs, have a local on-board, occupancy LED for testing, and also indicate when power was connected and “ON.” It also needed to be easy to build and low cost. Glossing over the multiple versions that were rejected, the Twint4 was built.

 

First Twint4 Design (Non Optmized)

The basic design uses a 1N5400 back to back diode pair to sense current flow to the block in both directions and turn on one of the twin pair of transistors (Q1 and Q2). A green 3mm LED is used to indicate power on, and a blue 3mm LED is used to indicate block detection. The detector operated with its own, independent 12 Volt power source. To detect the presence of a loco or car when the throttle was off, the detector injected a small current into the block via a resistor from its 12 Volt supply. So far, so good. However, the original breadboard showed some odd behavior.

 

My “power pack/throttle” was a “laboratory-grade” power supply. The detector “sort-of” worked. It would certainly detect the presence of a loco and allow movement forward and back. But it would not detect the loco with the power/throttle “off.” …Huh??  Undaunted, I found the last commercial DC throttle I once used, and tried it. I got slightly different results, this time with reduced speed in one direction, even though it did detect the loco now when the throttle was off. …What? …Huh?? I went back to search the literature and found 2 odd references. One was in Don’s 1980 article where he noted that with many solid state throttles the Twin-T may not detect a car in the block because the throttle voltage does not go to zero when the throttle is set to zero, and as such, cancels the bias voltage (from the DB connection on the detector). In a letter to the Clinic editor in MR October 1972 a “solution is proposed” using 4 diodes to “isolate” the power pack from the detector. I tried the 4 diode solution, and it changed but did not universally solve the problem.

Don was correct about the source of the problem. The small injected current interacted with modern regulation and momentum circuitry in later DC throttles. It really interacted with my lab power supply, the output of which is monitored by the supply, and internal circuitry is constantly active to forcibly maintain the voltage at the output under all circumstances – including overriding current from small block detectors! But wait – for decades this was virtually a standard for block detectors! I went back to my old DC throttle and turned off the momentum and braking. More change – the detector and loco behavior  was a bit better. Back to the archives of ancient goodies – I found one of the first DC power packs I ever had, hooked it up and… everything worked perfectly! Oh boy, oh boy, oh boy!! But why? Well, because in the 1950’s and 1960’s the most common throttle and power pack was an unfiltered, unregulated, DC supply attached to the track via a rheostat (a variable power resistor). There was nothing to interfere with the Twin-T. The Twin-T was a good DC block detector for its time. The more electronics you inserted into loco control, the more unpredictable the results without understanding the electronics. Fortunately for my friend, his throttles of choice were all old, but solid throttles, without complicated electronics, and the Twint4 detector worked great! Yeah!!

Along the way, I also read that Twin-T detectors were incompatible with DCC layouts – wrong! Without the bias resistor (the R2  510 Ohm ½ Watt resistor DB connection) this detector works well with DCC blocks, and only requires one rail to the block to be gapped. Please see the DCC hook up diagrams below.

The output of the detector (LOAD) is an open collector output which can be connect to a logic circuit, but make sure you use a pull up resistor connected ONLY to the Arduino/logic power (Usually +5 Volts). Remember that these were originally used for panel indications of track occupancy. The basic detector presented here will have LEDs flickering from dirty track and wheels and in block crossings. There is no signal conditioning and there never was without additional circuitry.

Low Cost Block Detectors for DC and DCC

Here is the updated Twint4 design with component mods for increased sensitivity and decreased power consumption:

 

Below you will find:

1. Details for fabricating small detector PC boards
2. A step-by-step guide for building the detectors
3. Detector connections to track and DC power packs/throttles
4. Detector connections to track and DCC base stations/boosters
5. A bill of materials (BOM) for the detectors with part examples from Digikey.com and alternate sources

You can leave off R7 and the green 3mm LED if you don’t want a “power on” indicator.

You can leave off R4, R6, Q4, and the blue 3mm LED if you don’t want an on-board block indicator.

You can substitute very low cost ($0.02 each) plastic package TO-92 2N2222A transistors for the metal TO-18 2N2222A’s, but the output current for the plastic package is 600ma max vs. 800ma for the metal TO-18. Also be very careful of the pinout order in the plastic package, they are reversed by certain manufacturers.

 

Metal or Plastic (Lower Cost) Transistors Can be Used

 

You don’t need to read schematics to use these. I designed a small PC board that anyone can order for themselves. It is a two layer board of 1.86x1.27 inches (47.27x32.16mm). Specific, step-by-step instructions for the build follow. The PC Board file named TwinT4.brd labeled with the original Twin-T component values can be downloaded here:  http://mrhpub.com/files/users/geoffbfiles/TwinT4.zip

An updated board with updated component labels named TwinT5.brd can be downloaded here:

http://mrhpub.com/files/users/geoffbfiles/TwinT5.zip

 

The PC Boards can be ordered from this board fabricator:   oshpark.com.

OSHPark offers a public service fabricating very high quality, low cost PC boards in low quantities. Set up an account at OSHPark (really easy) and upload the board file you want to fabricate and specify the quantity (always in multiples of 3—their rules) unless you order the large quantity slower run boards. They accept Paypal. I have no vested interest in OSHPark – I’m just a satisfied customer. Please feel free to use whatever fabricator you know. The bare boards are shipped “panel-ized” so break or cut them apart.

 

New update: It looks as though Model Railroad Control Systems will be stocking the bare board as of December 2017 or so. Look for them here on the "Community Page" at  http://www.modelrailroadcontrolsystems.com/community-designs/  If any other source becomes available I'll post it here. I have no stake in any of these sources.

 

I want to thank my friend Vince, for encouraging me to explore this path and solve an old problem for me, and a newer one for him. These are relatively easy to build and could easily be used with DC, DCC and AC power. If you hand wired these on perf board with plastic transistors it would be possible to build a block detector for less than $4.00 per detector. I hope some other modelers can make use of these too.

Have fun!

Best regards,

Geoff Bunza

 

Let’s Build Some Block Detectors

Follow the following build diagrams and solder the components to the board. Use a low temperature soldering iron or a low wattage soldering iron for these.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Original Detector and Detector with Updated Labeling

 

 

DCC Detector- Layout Connections

 

DCC Detector- Layout Connections

 

For good prices for quality surplus electronics you can use:

https://www.allelectronics.com/

for transistors, diodes, LEDs, switches, relays, and many other components

Digikey.com    and  Mouser.com   for brand new high quality components

http://www.surplusgizmos.com/

for LEDs especially these high brightness 3mm green LEDs:

http://www.surplusgizmos.com/search.asp?keyword=green+3mm+led&sortby=0&catid=171

but for many other components too.

Ebay.com offers very low prices but be wary of some low quality parts sometimes offered.

 

Twin-T4 BOM	Need Per Board	Digikey Part Number	Manufacturer Part Number	Description	Part Cost	Board Cost
1N5400	2	1N5400GOS-ND	1N5400G	DIODE GEN PURP 50V 3A	0.19	0.38
2N2222A	5	2N2222ACS-ND	2N2222A	TRANS NPN 40V 0.8A TO-18	0.40	2.00
510 Ohm 1/2 W	1	510H-ND	CFR-50JB-52-510R	RES 510 OHM 1/2W 5% AXIAL	0.10	0.10
1000 1/4W	3	1.00KXBK-ND	MFR-25FBF52-1K1	RES 1.0K OHM 1/4W 1% AXIAL	0.10	0.30
10,000 1/4W	3	10.0KXBK-ND	MFR-25FBF52-10K	RES 10K OHM 1/4W 1% AXIAL	0.10	0.30
Blue 3mm LED	1	LTH3MM12VFR4600-ND	LTL1CHTBK5	LED BLUE CLEAR 3MM ROUND 12000mcd	0.25	0.25
Green 3mm LED	1	732-5012-ND	151033GS03000	LED GREEN CLEAR 3MM ROUND 15000mcd	0.25	0.25
Screw Term 5mm	1	732-10959-ND	691137710006	CONN TERM BLOCK 5.0MM 6POS PCB	1.27	1.27
PC Board via OSHPark.com	1	TwinT4.brd	High Volume Build	OSHPark.com Board Fabrication	2.35	2.35
				Single Board Cost		6.97

From another thread but pertinent.

Hi Dr. Geoff,

I just got my PD6Lboards from OSH Park.  They're great and great people/system to work with too!  Thanks for the tip!

Enjoying!

Nick

Thanks!

Geoff,

Thanks for all the things you do to support our wonderful hobby! To dig into this problem and take the time to finally wrestle it into submission and then give a working design and detailed instructions on how to build it is a testimony to who you are. 

While I do not have your technical skills you continue to provide me with the inspiration to do what I can to help out other model railroad hobbyists whenever I can.

Ross A.

@Nick and Ross

Hi Nick and Ross,

Many thanks for the kind words. I hope many modelers can enjoy using these in any form, and make some really great layouts.

Have fun! 

DCC on DC

This is a really neat project, I still use straight DC. So, with DCC units running on DC I just leave the voltage on at around 6volts or so since I don't want my units to go through the start-up sequence over and over.

I take it that this would enable the block detector to function almost always, or have I misread this?

I'll buy a few of these and try building them if that's the case.

@chepe

Hi chepe (name?),

Yes, with your 6V applied to the track the detector should work. Here's the caveat: I tried exactly what you are describing and it did work, but the current draw from a decoder at standstill on a DC track could be very small, especially if it is not a sound decoder. By my measurements it needs to draw somewhere between 7-9 milliamps (0.007 Amps) to trigger the detector. It's not much but a DCC decoder operating on a DC track without sound might draw very little. I would try one out before you instrumented your whole layout and committed to a larger investment. There aren't many simple detectors for DC, but this is one.

Have fun! 

Geoff, Thanks for the deeper

Geoff,

Thanks for the deeper explanation, to date I only have three with regular decoders, the rest are sound DCC and a gazillion straight DC that haven't been converted yet.  Your advice about trying one first is taken seriously, will do.

Chepe

Geoff, Thanks again for all of us

Geoff,

    Thank you once again you are helping many of us with solutions that can be used with JMRI as well that you from all of us.

Mike

@Chepe and Mike

Hi Chepe and Mike,

You are most welcome. 'Hope can make use of this for your great modeling.

Have fun! 

Some Questions

Hi Geoff, thanks for this!  I like the low cost, build it myself approach.  I do have some questions though.

1. Is the only difference between the two circuit versions the values of the resistors?  Or did my old eyes miss something important?
2. For the Load terminal, how much current can it handle, i.e. how many LED's can I attach?  And are there any limits to the wire length between the board and the LED's?
3. Likewise, are there any limits on cable length between track and detectors?
4. In either case, if I have to run wire for any significant distance, should the be parallel or twisted pair?
5. You reference R1 as the bias resistor, I assume you actually mean R2.  And to be clear, that resistor is not needed when building for DCC layouts, correct?
6. To help me understand the circuit, what does "DB" stand for?  ("Detection Bias"?) And what is the "C" terminal for as it never seems to be used?
7. I assume that nothing special needs to be done when using multiple DCC boosters?  Ditto for DCC circuit breakers:  these just connect in line between the one rail and the output of the boosters or breakers?
8. I'm looking at detecting only in a staging yard.  Can the rest of the layout be a single undetected block?   Will the 1N5400 diode pair be able to handle the current draw from the several engines running through them?  And, in DCC, why is it needed?  Just to keep the voltage on the detection rail uniform?
9. Your commentary suggests that there are better detectors out there for DCC.  Are any of them as easily and inexpensively built as this that you would recommend?

Thanks again!

Richard Kurschner

@Richard re:Many Good Questions

1. Quote:

   Is the only difference between the two circuit versions the values of the resistors?  Or did my old eyes miss something important

The plastic cheap 2N2222A transistors max out at 600ma, so adjust down. I use high efficiency, bright, cheap LEDs at 8000-15000 mcd which I often use a 10K dropping resistor to run the LEDs at less than 1ma each! (like the green 3mm LED I used for the on board power indicator. I doubt you are going to run into a length problem to a panel. The biggest issue would be the voltage drop due to wire resistance to the panel LED. Try keeping your panels within 100 feet of your detector!! and you should be fine. Please remember these were originally designed to be simple detectors for panel occupancy indicators. Speed to the LED is also not an issue.

'Hope this helps. Have fun! 

Comment on Twin-T

I hate to be the guy who looks at something that's operated for decades and says it can't possibly work, but there's a feature there which I'd call poor electronic design. It's the way that the transistors Q1/Q2 have their base-emitter connections in parallel with D1/D2. The full track current runs through that circuit, and how can anyone know what the split will be of current flow through the transistor bases versus the diodes? A p-n junction has a forward voltage drop of "about" 0.6V, but it's not something to rely on. I'd be more comfortable if the transistor bases were protected by a resistor each. And then, is it guaranteed that there's enough voltage drop in the diodes to turn the transistors on? Apparently it works, but I'd want to put 2 diodes in series in place of D1 and again for D2. Yes, there'd be extra voltage loss and cost, which are not good things! Perhaps it's an insult to the memory of Lin Westcott, but I'd look at an op-amp circuit for this job.

For use on DCC, you might get away with a simpler design, because you can be certain that both positive and negative current flow will occur, so you only need single-ended detection; it would be intermittent, but it could still light a lamp. And anyone who wants to interface the detector with other electronics could consider an opto-isolator in place of (or in addition to) the LED that's called "Load". That way you'd isolate the track-connected system from any other devices that were in use.

@John re: Twin-T design

Hi John,

One last thing as pointed out in the blog, this detector in its modified form works just as well as my preferred "non-invasive" coil block detectors for DCC with simple, cheap components -- who would have thunk (sp) it !!

Have fun! 

DCC Polarity

Hi Geoff!

My Digitrax command station does not have a common rail designation.

Your electrical drawing shows a DCC rail common to the 12v ground.

I cannot mix DC with DCC, it causes my DCS240 to not power up.

Could the Command station GND be common with the 12v GND?

I have built 3 of these & tested them, they are ready to go on my layout.

Jay

@Jay re: Common in DCC???

Quote:

Your electrical drawing shows a DCC rail common to the 12v ground.

I cannot mix DC with DCC, it causes my DCS240 to not power up.

Could the Command station GND be common with the 12v GND?

Good catch Jay, but this is more a matter of wording/vocabulary but it does require at minimum an explanation:

When you wire a DCC layout the most simplistic was of thinking about it is to consider the rails the DCC bus, but when you get to signaling and block detection via current sensing, you need to divide your track into sections or blocks. Some modelers even gap both rails, but for current sensing you really only need to gap one rail side for block isolation (gaps at both ends). The reason is that you want to measure any current going anywhere into the block. If you do this consistently around your layout then the ungapped rail side could be cosidered a "common" connection (not ground) for your DCC wiring.This is a matter of terminology, and has nothing to do with a common ground sometimes mentioned in DCC manuals and instructions. Sorry, but engineers can actually have multiple "common" connection points in a large design. I should have drawn this distinction more clearly. Thanks for bringing this up.

 

I also want to take this opportunity to make absolutely sure that the 12 Volt Detector power supply be completely independent of your DCC Power supply. There should be no other connections other than the connection at COM on the detector board. The DCC signal to the block goes though the 2 back to back detector diodes as it feeds each block rail. Since all the COM connections are tied together for multiple detectors, it was labeled COMmon in the original design. I left it that way. Once you decide which side of the DCC bus is connected to COM you need to be consistent throughout your layout. I have actually tried these on my Digitrax DCC system and it does work just fine.

 

I hope this is a bit more clear. If not please ask more questions.

Have fun! 

Best regards,

Geoff

Question

Hi Geoff,

Thanks for the response.

From your drawing it looks like 12v supply positive goes to PWR & it's ground goes to COM.

Also, it looks like detector side of the DCC Signal also goes to COM.

Will that cause a voltage issue riding on COM? Pushing that GND to track voltage?

It does not compute in my Techie brain!

Jay

@Jay re: Question

Hi Jay,

'Hope this helps. Have fun! 

Issue

Hi Geoff!

One of the ones I built is lighting the blue led all if the time.

It powers up fine otherwise.

I think one of the transistors is bad.

Can you narrow it down for me?

Thanks!!

Jay

@Jay re: Blue Led lit

Hi Jay,

If the blue train detected LED is ON it means that transistor Q4 is turned ON, which normally would imply that transistor Q3 is turned OFF.. I would first check if transistor Q5 is also turned on by connecting a LED in series with a 1K to 10K resistor to LOAD and PWR with the Anode (Plus side) of the LED to PWR. If this test LED also lights, it means something is either forcing Q4 to turn ON or forcing Q3 to turn off. See if there are any solder bridges or incorrect components around Q4 and Q3.

Next, with nothing on the track,temporarily connect DB to C and see if the blue LED turns off. If it does not turn off, the Q4 may be shorted (bad) or Q3 may be stuck open (bad). Check to make sure nothing is connected to terminal "T"

Let us know how you do. Have fun! 

Twin-T 5

What is the best way to trigger a 12 volt relay using your Twin-T?   The relay would have contacts for signals and for turning power off to a track section before a red signal light.   I suspect that a transistor off one of the LEDs would control the relay (12 volt).

I used Linn W's old Twin-T back in the 1970s to run a relay that ran the signals, block occupancy lights and a track sections power before the signal.  I could run two trains in six blocks for hours with not one colliding with the other.  I could go to dinner and returned to find the trains still running.  When the signal turned green power was restored to the dead track.. 

I want to create the same on my new layout. 

Is there a way

@buddy76 re: Use with 12 Volt Relays

Have fun! 

When does the detector trigger?

Geoff,

Thanks again for yet another great tutorial. 

In your experience, does this detector trigger as the first axle of a locomotive enters it, or once all axles of the first powered unit enter the detected block from the adjacent one?

I ask because I would like to use these to precisely align station stops for multiple different types of DCC transit vehicles and hope not to use IR due to some limitations on space.

Thanks! 

AJ 

Los Angeles

@AJ

Hi AJ,

The detector senses electrical current going through the isolated rail of the block. I was very careful with the wording of the last sentence. The isolated rail is the RED rail in my diagrams. Even though the BLACK rail can go through block selector switches, all the blocks for DC control are common to one power pack when used. In the DCC connection BLACK represents the common rail. I am being so specific because the detector needs to see current flowing (in any direction) from the BLACK common rail to the RED isolated rail of the block in question.

Now for the trouble: Think about a 2 truck diesel or a steam loco where power pick up is separate (left from right rail) by truck, or by loco and tender. This occurs when the loco drivers only pick up the right rail, and the tender only picks up the left rail. Depending of the orientation of the loco relative to the isolated rail, your loco could enter the block the length of the loco and not be detected until the tender or the second truck entered the block too. Example, loco pickup in on the right, but the isolated rail is on the left-- there will be no loco detected until the tender enters the block.

Sorry to beat this into the ground, but it is a problem hat is not intuitively obvious to many modelers. If you are careful as to how you gap the isolated rails you can fire the detector to stop your train at a specific point. But, be careful about travel in the reverse direction. It can be problematic depending of the length of the loco and the length of the block. As you might guess, locos that have all wheel pickups do not have such an exaggerated problem.

I hope I made this clear. Sorry for the ramble. Have fun! 

When does the detector trigger?

Thanks for the reply Geoff. So with all-wheel pickup, it would not trigger until all the interconnected axles entered from the prior block?