Join Now
International Automotive Technicians Network
VW Diesel No Start and Cuts Out at High RPM
Posted to Technical Discussion Forum on 11/22/2008 14 Replies

Do you guys remember this one?

Well I've got an update, and fix for it. I'll warn you ahead of time. This is a long one. Well here goes:

Subject vehicle: 2004 VW New Beetle 1.9 "C" BEW diesel engine

We do some diagnostic work for a used car lot in town. They asked me to look at this New Beetle with a CHECK ENGINE light on, and we agreed with stipulations. The reason for the stipulations were that we don't normally work on too many foreign cars, and even fewer VW's. So the agreement was to do a baseline check and get back to him. His complaint was an intermittent "cranks but won't start". He stated that if he turned the key on, and the "glow plug light" was flashing, then the car would not start. But if the light came on for a little while then went off, it would start and run perfectly. Initial testing revealed trouble codes P0727 (engine speed input circuit no signal), P0674 (cylinder #4 glow plug circuit open), P0673 (cylinder #3 glow plug circuit open), and P0672 (cylinder #2 glow plug circuit open). We proceeded to test the glow plugs, and the car had only one good plug. The other 3 had various amount of heater element missing. The customer was informed of the possibilities of additional problems and possible repeat overheating and we replaced the glow plugs. We disregarded the CKP code due to a comment on iATN that any time the engine cranks excessively, that code will set. After replacing the glow plugs, the car did not start right away, but before we could check anything else, it started and ran and continued to work normally for a couple of days, so he picked up the car only to return a couple of days later with the same complaint. This time the only code stored was the P0727 (engine speed input circuit no signal). We found the glow plugs to be working normally, using and amp probe, so we began to look elsewhere. A simple resistance test of the CKP sensor revealed it to be partially shorted according to manufacturer's specifications, so we waited several days for a new OE sensor. The OE spec is 1k to 1.5k ohms. Our old sensor read about 650 ohms. We installed the sensor, and the car did not start for several minutes. We spent some time trying to decide what to test next, and a few minutes later the car started normally. It would not act up again for several days, so the customer drove it again, only to return with the same complaint, as well as a new one. Now it also cuts out intermittently above about 3000 rpm. Since it didn't do that before the sensor, and we didn't notice it after the sensor, we thought that maybe we were going to get a break and be able to find something more consistent. At this time, I tried to get my customer to go to the dealer, but he would have none of it and wanted me to try to fix his car. I enlisted the help of the guys on iATN, but apparently this is not a common engine and I could get no concrete answers to my questions. We did get a tip about the electric lift pump located in the fuel tank causing an intermittent no start whenever the pump did not run, so we tested ours only to find that it was dead. We talked to our customer and he said he would only buy it if that would fix his problem. I told him that there was no way I could confirm that it would be all of his problem, but that it was indeed dead. He was afraid of getting in much deeper, since he was way over his head already. Since I could get no definitive answers from my friend on iATN, I had a heart to heart with the customer again. He agreed to allow me some testing time to try and confirm the lift pump issue. By now, the car was dead and would not start for about a week while we were waiting on him to make up his mind about that $500 pump.

I began my testing on a car that I knew little to nothing about by starting with the basics. What we wanted to know was whether replacing the pump in the tank was a necessity. Doing a little research I discovered that this engine has an electric lift pump in the tank that creates pressure to flow through the fuel filter and into another pump, called a tandem pump mounted on the end of the cylinder head, driven by the camshaft. The tandem pump is basically a high pressure pump mounted to a vacuum pump. Since a diesel doesn't create any usable vacuum, it is common for them to have an engine driven vacuum pump. Since I have seen other diesels run and start just fine with dead lift pumps, I decided to test this one to see if it would run with just the tandem pump (high pressure pump) sucking fuel from the tank instead of being fed pressure from the lift pump in the tank. My first test was to apply a vacuum to the line going into the tank. I could not pull a stream of fuel, but a whole lot of air. Tracing it back, the problem was internal to the lift pump in the tank. At this time, thinking I might have a bad line in the modular assembly, we pulled the fuel lines and discovered the fuel seemed to be a little thick, similar to vegetable oil instead of diesel fuel. I removed the lift pump and took a sample of the black liquid that should have been clear fuel. Because we still didn't know if this thing would run with a solid column of fuel, I put a piece of pipe in the fuel line to give me a solid column and proceeded to vacuum bleed the line to the engine. After a little cranking, the engine still did not start. I was beginning to think that the pressure from the lift pump might be required, to get this thing to start. Also, I wasn't sure how much air had been ingested into the injectors and how long it would take to get it out.

At this time, I installed a pressurized container of diesel fuel to the tandem pump inlet to simulate the lift pump. I didn't know how much pressure was needed, but since my interpretation of the manufacturer's test was that the lift pump should produce .5 bar (about 7 psi) I started there. I decided to let the fuel run for a while in case there was a bunch of air in the injectors since I had already verified that the inlet was not full of fuel. The car still did not start. At this time, I decided to tap into the high pressure port with my fuel supply. This factory port measures rail pressure, and the only spec we had was with the engine running. Since we already had a gauge in there previously when the engine was running, we knew that it would idle somewhere around 60 psi and raise with rpm up to about 100 psi. I put about 60 psi on it, and the car started immediately. I tried revving it up at this time to see if the cutting out at 3k was still there, and it was. So, when it ran a little, I shut it down, and hooked the feed line back up to the tank which still had the solid tube in lieu of the lift pump. Since the engine was now running, and it was late, I cleared all the codes and let the car run for a while. Remember, what I was trying to determine was, whether or not this thing would run without pressure from the dead lift pump. I went home for the evening with happy thoughts, and the next morning, the engine would not start. It was at this time, that I called my customer and told him that whether or not this pump fixed this car, there was no way that it would run with this one, since it was obviously sucking air. At that point he found a cheap pump from the local Napa store, and when it showed up, we discovered that it weighed half what the factory one did, because there was no pump in it. It was just a sender, so he elected to wait a week to get the OE pump.

When I got the OE pump, I proceeded to test it with the same vacuum test that I used on the old one. The new OE pump would not allow a solid column of fuel to be drawn to the engine either. After the new pump was installed, the fuel tank was drained of all the black liquid, and the system was bled out and running, we still had to deal with the cutting out problem at higher rpm.

We knew that it didn't do that before. We also knew that the customer said it didn't do that before the new CKP sensor. We also knew that the new OE CKP had the correct resistance. After test driving with a VAG COM scan tool loaned to me from a friend on iATN. Thanks Tony. I verified that when the engine was cutting out, the rpm on the scan data was faltering. Rather than just reinstall the old CKP I decided to do some testing. I remembered previously when the engine was not starting, we had the scope leads installed in the CKP connector, and I tripped over the leads, jerking them loose. Almost immediately afterwards, the car started and ran. I also had discovered that each wire on the sensor circuit had about a 2.5v dc bias. So I used this to my advantage to do a wiggle test of the harness, but everything was perfect. So I grabbed some patterns and removed the new OE CKP and reinstalled the original (low resistance) one. A comparison of the two sensors revealed a little different configuration and one other thing interesting. The sensors stuck together magnetically. I thought that it was possible the reverse polarity of the magnet might be the reason for the driveability concern. Is it possible the signal is backwards? So I compared the two only to discover that they were identical. The only difference was that now the car ran great. I had installed the original "bad" sensor.

Trying the car over the next several days revealed several things that gave me answers to some of my questions. When the car was running, and the lift pump was unplugged electrically, the car would die after it ran for several minutes, presumably when the solid column of fuel from the tandem pump to the tank turned into a mixture of air and fuel. If the lift pump was disabled and the engine shut off before the fuel column was used up, the car would restart, at least within a few minutes. When a solid line to the tank was installed, the car would not restart after setting over night. When the lift pump was dead, and the old CKP was installed (the one that made it run good) no amount of cranking would set the P0727 code, or any other one. Granted, I didn't crank it enough to burn up the starter, but I did try it numerous times and even some very long crank times. When the car was not starting with the old sensor, the code would set easily. But here is the kicker, even though it was setting the code, it was still toggling the injector control circuits. The CKP low amplitude was setting the code cranking, but was not preventing it from starting.

I believe that we started out with an intermittent dead lift pump, and bad glow plugs. Then installed a defective new OE CKP sensor, and were then fighting air in the fuel system. It's been about a month now, and the car runs great. Hope this helps someone in the future. I'll follow up with a couple of quick tests that I discovered in my venture.

If you get one of these with a no start, tap into the rubber lines at the top front of the engine going to the tandem pump. You should have fuel pressure of about 7-10 psi with a 2 second run KOEO, and anytime you are cranking or running. If the lift pump does not run, fix that first. It is essential. Then, on the LH end of the head there is a connector that plugs into the head the feeds the fuel injectors. The injectors are pumped up by the cam and controlled by the PCM. Use an amp probe on the brown wire to see if the injectors are being toggled. Ours were being toggled even though we had a no start. You do not need the CMP for this thing to run. Disconnecting it did not turn on our CEL, but the engine did die if we accelerated to about 3000 rpm. Our tandem pump would put out about 50-60 psi cranking, with a no start, and with the engine running. It would ramp up to a little over 100 psi as rpm is raised. Just because the resistance is wrong or right does not mean the CKP sensor is good. It's all about signal amplitude. Ours ran great at low rpm and started good once we got the fuel system fixed, but it would cut out and or miss at higher rpm consistently. It got a little better when warm, and almost ran good at normal temp. The CKP has three wires, but is a two wire variable reluctance sensor that has 2.5v dc bias on each signal circuit and one shield wire. When tapping into the CKP signal, the sync portion of the signal from the brown wire goes up first, and the white wire goes down first.

Here are some shots that I grabbed for your entertainment:

New OE sensor (BAD) at idle

New OE Sensor (BAD) around 3400 rpm

Good CKP (too low resistance) at idle

Good CKP (too low resistance) around 3400 rpm

CKP white wire and injector amps cranking

CKP white wire and injector amps cranking #2

Good CKP brown wire at idle

New OE sensor white wire (BAD) at idle with cmp

Fuel pump on bench

Message to everyone

Engine orientation

Tandem fuel pump

Fuel Sample

Fuel Sample in Glass container

Good and bad new CKP sensors

I'll have to get back to you on which sensor is which. It's Saturday and my tech is gone.


Scott from Missouri

Files Referenced:

thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail thumbnail

14 Replies Received (View Replies)


car Vehicle Data

2004 Volkswagen Beetle GL 1.9L Cuts Out, No Start

Engine1.9 L
Trans5-speed Standard
SymptomsCuts Out, No Start