DIY Stay Alive Advice Needed

[JSModels]

I'm looking to make a stay-alive for my new Hornby Ruston 48DS so I can run it without the match truck.

 

What I want to get from it: I want to achieve reliable slow running on my shunting puzzle layout, so need to overcome the occasional bad contact. I keep the track reasonably clean, but I'm not religous about it. Having said that, I don't need it to run for 10+ seconds without any power applied like some videos I've seen! I'd like to use as little space in the cab as possible (definitely keeping it below the windows) and keep the chip where the socket is above the motor under the bonnet.

 

I've bought a Zimo MX617N chip to replace the Bachmann one I temporarily installed, and want to build my own stay-alive to connect to it, but I have questions:

 

1. Capacitor Voltage: are 16V capacitors sufficient, or should I go for 20V or 25V? If I go for a higher voltage, it does limit the ultimate capacity I can have. Some commercially available stay-alives/kits seem to be 16V, others are 25V, so I'm really not sure. The Zimo manual states to only use 16V if the track voltage is lower than this, but I've no idea what my controller outputs (and the manual doesn't state it). I'm assuming that due to the way DCC works, simply measuring across the track with a multimeter won't be helpful!

 

2. Capacitance: What total capacitance should I be looking to achieve? I could use multilayer ceramic capacitors (such as these) in which case I could use 16 of them in a 4x4 array & have 1,600uF, which would only take up a 10x10x4mm space. But would that be overkill for what I need (see above)?

 

3. Capacitor Type: Are the ceramic ones the best option for me, given the above? The cost of them, whilst not prohibitive, is certainly a factor. Something like this polymer aluminium electrolytic capacitor would give me 680uF and is only 8mm dia x 12mm long, so would still fit easily within the space (maybe even 2 of these?) And it's WAY cheaper. Then there's tantalum...

 

4. Control/Charging Circuit: The Zimo documentation shows a charging resistor (68 ohms) and a diode, but also a slow-discharge resistor (3K3) and a choke. Some circuits I've seen omit the choke, and some use a zener diode, which isn't shown on this circuit?

 

Basically, I need someone to walk me through this like I'm an idiot...

 

Jonathan

 

[Nigelcliffe]

1 Read my blog, going back over years.   Sorry, you'll have to dig around in it, but there is quite a lot in there.  Link in sig file at bottom of post.

 

2 Forget the ceramic capacitors you linked to, they don't store energy very well at full voltage.  Tantalums offer much better effective energy for the volume occupied in the model.   See my blog entry on them.

 

3 Track voltage, find way to measure it.  Simplest is measuring the DC voltage over decoder positive and decoder ground.   
      You can't use a cheap multi-meter to measure the DCC track voltage directly. 

 

4 Circuit - minimum which works reliably is charging resistor plus diode to send full power back to decoder (RH page of diagrams you quote).   The choke is only needed to re-load firmware into non-sound Zimo decoders, not something most people will ever do.  The discharge resistor only needed for very large capacitances (say well over 10,000uF).    

 

5 Capacitor voltages,  you'll find this is a trade-off.  16v rated stuff will work, but you're near the limit on any track voltage.  But, 16v rated stuff is a LOT smaller per unit energy than 25v rated.   So, need safe solution to use 16v rated parts in tight spaces. 

     5a  - know your track voltage, and if you know its well below 16v you're probably going to be OK.

     5b -  use the Zener diode (shown in some circuits) to limit the voltage over the capacitors (secondary protection).  The charging resistor limits the current through the Zener if it comes into play.

     5c  -  use the Zimo SACC16 charging circuit if you have the space.   This is my preferred solution now, unless I'm really stuck for space, its cheap (£10), and it has effective 16v component management, so safe on any track voltage.    YouChoos.co.uk  have quite a lot of documentation on the SACC16.   

 

6  Capacitance.  As much as you can squeeze in, at least until you get to well over 2,000uF.   Around 600uF there is an effect, but it is a really tiny amount of wheel movement - barely a twitch. 

 

 

7  Go to WorkShop Wise near Morpeth, Northumberland, next Saturday, bring your bits and I'll talk it all through with you.

 

 

 

- Nigel

 

 

[RobjUK]

I've just got a batch of these to use for compact stay-alives. 0.47F at 6V so three in series give 16V with some tolerance and roughly 150,000uF

Each one is two 6mm diameter caps held together with heatshrink sleeve, 14mm long.

https://www.digikey.ca/product-detail/en/avx-corporation/SCMQ14H474PRBB0/478-11305-ND/7595421

 

The ones in that series are specifically rated for high current use, unlike most of the supercaps on ebay; 0.59A for that exact one, plenty for a small loco that's already moving.

 

If you can fit the three in different locations that's fine as long as you series them up. Add eg. a 1K resistor across each to keep the voltage equally divided, then use the diode + resistor with 15 or 16V zener as Nigel says.

 

[JSModels]

Nigel,

 

Thanks for your help - I was hoping you'd see this! I've read quite a bit in your old blogs, including where you said you had issues with the ceramic capacitors (you've helped me out with chip recommendations over on the n gauge forum before).

 

Thanks as well for the info about the SACC16 - despite a lot of searching over the weekend, I didn't find any links to this, and didn't know it existed! It definitely solves a lot of problems.

 

With the add-on tantalum caps, it maxes out at 1100uF, so might try and squeeze one of the supercaps in there.

 

Just a thought - does it matter if you mix capacitor types? For example, if I fitted one with the tantalum caps, and found it was still occasionally stalling, could I then add a supercap in parallel with the tantalum? Or does mixing the types cause issues?

[Nigelcliffe]

JRB

there isn't really a "max out" on the SACC16.  Yes, if one fits 5 Zimo-badged 220uF Tantalums its 1100uF, but there are larger Tantalums available (YouChoos list a lot before you get into searching electronics suppliers, including some 6800uF and larger "brick" capacitors).   And, there is nothing to stop you wiring additional units in parallel to the five contacts provided - this gives more space options, rather than making one single rectangular package.

 

Mixing capacitors, in parallel its generally OK.   Mixed arrangements in series it gets complicated quite quickly (and into areas where I'm not confident), so for my thinking, any which go in series need to be (a) all the same and (b ) have balancing resistors over each capacitor (as RobjUK said) to hold the voltages equal over the series.   
I don't know how some of the chemistry's quoted work out - quite a few capacitor designs are not suited to the high currents needed for a stay-alive.  Even the 0.59A RobjUK mentions may be marginal on a Zimo and a fairly small loco/motor if it attempts to crawl the loco away from a dead spot, which the decoder will attempt to do. That results in maximum current draw for the motor, starting from standstill and operating at low speed, until the capacitor is drained.

 

 

[JSModels]

Well, I've just ordered one of the youchoos 'LifeLink' ('own-brand' version of the SACC16) along with a 17,000uF supercap.

 

It's probably way overkill, but it's the same physical size as the 6,800uF supercap - not sure how that's possible, but hey!

 

JRB

[JSModels]

Last night I ran in the Ruston on another member's layout, and then replaced the temporary chip I'd installed with the MX617N. The supercap and control board haven't arrived yet, but I did manage to solder the wires for it to the chip.

 

A few minor CV changes later and, even without the stay alive, it's running superbly. It did still stall in a couple of places, and was slightly stuttery in others, but once the stay alive is fitted that should be cured.

 

Forgive the lack of cab roof, and the wires sticking out!